DL_102_Primitive_neurons.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "DL 102 - Primitive neurons.ipynb",
      "provenance": [],
      "collapsed_sections": [
        "7Ju44HNZE-5g",
        "MfmGSVAdFMV5",
        "bZNZ86t_Qbyv",
        "_rWyQyUWYrZs"
      ],
      "authorship_tag": "ABX9TyNe35/UOBb/t4iNgQohMucr",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/4deepprk/mp_neuron/blob/master/DL_102_Primitive_neurons.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "7Ju44HNZE-5g",
        "colab_type": "text"
      },
      "source": [
        "# Loading dataset\n",
        "\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TqnOncSC-e3X",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# MP neuron model\n",
        "# Using real world dataset breast cancer\n",
        "\n",
        "# ML package for machine learning techniques and data sets\n",
        "import sklearn.datasets\n",
        "import numpy as np # numerical computation"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "m3uk-_4k_RYp",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# real world breast cancer dataset that is loaded using scikit learn\n",
        "breast_cancer = sklearn.datasets.load_breast_cancer()"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "DxRO6Vdm_qmb",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# X as features and Y as class labels\n",
        "# Binary classification problem (malignant or benign)\n",
        "X = breast_cancer.data # .data gets features of breast cancer data\n",
        "Y = breast_cancer.target # target name gets the labels"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "FeI1UaPAAVty",
        "colab_type": "code",
        "outputId": "4109acf7-28b5-407c-8340-166e6264c95d",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        }
      },
      "source": [
        "# 2 dimensional array or matrix or feature set or series of data items\n",
        "# each row contains features of sample\n",
        "print(X)\n",
        "print()\n",
        "# 1 dimensional array or vector or label set\n",
        "print(Y)"
      ],
      "execution_count": 197,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[[1.799e+01 1.038e+01 1.228e+02 ... 2.654e-01 4.601e-01 1.189e-01]\n",
            " [2.057e+01 1.777e+01 1.329e+02 ... 1.860e-01 2.750e-01 8.902e-02]\n",
            " [1.969e+01 2.125e+01 1.300e+02 ... 2.430e-01 3.613e-01 8.758e-02]\n",
            " ...\n",
            " [1.660e+01 2.808e+01 1.083e+02 ... 1.418e-01 2.218e-01 7.820e-02]\n",
            " [2.060e+01 2.933e+01 1.401e+02 ... 2.650e-01 4.087e-01 1.240e-01]\n",
            " [7.760e+00 2.454e+01 4.792e+01 ... 0.000e+00 2.871e-01 7.039e-02]]\n",
            "\n",
            "[0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
            " 1 0 0 0 0 0 0 0 0 1 0 1 1 1 1 1 0 0 1 0 0 1 1 1 1 0 1 0 0 1 1 1 1 0 1 0 0\n",
            " 1 0 1 0 0 1 1 1 0 0 1 0 0 0 1 1 1 0 1 1 0 0 1 1 1 0 0 1 1 1 1 0 1 1 0 1 1\n",
            " 1 1 1 1 1 1 0 0 0 1 0 0 1 1 1 0 0 1 0 1 0 0 1 0 0 1 1 0 1 1 0 1 1 1 1 0 1\n",
            " 1 1 1 1 1 1 1 1 0 1 1 1 1 0 0 1 0 1 1 0 0 1 1 0 0 1 1 1 1 0 1 1 0 0 0 1 0\n",
            " 1 0 1 1 1 0 1 1 0 0 1 0 0 0 0 1 0 0 0 1 0 1 0 1 1 0 1 0 0 0 0 1 1 0 0 1 1\n",
            " 1 0 1 1 1 1 1 0 0 1 1 0 1 1 0 0 1 0 1 1 1 1 0 1 1 1 1 1 0 1 0 0 0 0 0 0 0\n",
            " 0 0 0 0 0 0 0 1 1 1 1 1 1 0 1 0 1 1 0 1 1 0 1 0 0 1 1 1 1 1 1 1 1 1 1 1 1\n",
            " 1 0 1 1 0 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 0 1 0 1 1 1 1 0 0 0 1 1\n",
            " 1 1 0 1 0 1 0 1 1 1 0 1 1 1 1 1 1 1 0 0 0 1 1 1 1 1 1 1 1 1 1 1 0 0 1 0 0\n",
            " 0 1 0 0 1 1 1 1 1 0 1 1 1 1 1 0 1 1 1 0 1 1 0 0 1 1 1 1 1 1 0 1 1 1 1 1 1\n",
            " 1 0 1 1 1 1 1 0 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 0 1 0 1 1 1 1 1 0 1 1\n",
            " 0 1 0 1 1 0 1 0 1 1 1 1 1 1 1 1 0 0 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 0 1\n",
            " 1 1 1 1 1 1 0 1 0 1 1 0 1 1 1 1 1 0 0 1 0 1 0 1 1 1 1 1 0 1 1 0 1 0 1 0 0\n",
            " 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 0 1 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1\n",
            " 1 1 1 1 1 1 1 0 0 0 0 0 0 1]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "nAG5pw2mAZ8L",
        "colab_type": "code",
        "outputId": "06e2dc4c-fa38-4e13-937f-3382bc2077a9",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        }
      },
      "source": [
        "# 569 rows and 30 columns with 569 labels associated with it\n",
        "print(X.shape, Y.shape)"
      ],
      "execution_count": 198,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(569, 30) (569,)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "zgRFj3AiBCsH",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import pandas as pd"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "s3DdWM-5Baob",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Reading data as dataframe instead of reading as array\n",
        "data = pd.DataFrame(breast_cancer.data, columns=breast_cancer.feature_names)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1IOMyDfSCD18",
        "colab_type": "code",
        "outputId": "2fd8653f-7017-4417-a7cc-488401b0cf35",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        }
      },
      "source": [
        "print(data)"
      ],
      "execution_count": 201,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "     mean radius  mean texture  ...  worst symmetry  worst fractal dimension\n",
            "0          17.99         10.38  ...          0.4601                  0.11890\n",
            "1          20.57         17.77  ...          0.2750                  0.08902\n",
            "2          19.69         21.25  ...          0.3613                  0.08758\n",
            "3          11.42         20.38  ...          0.6638                  0.17300\n",
            "4          20.29         14.34  ...          0.2364                  0.07678\n",
            "..           ...           ...  ...             ...                      ...\n",
            "564        21.56         22.39  ...          0.2060                  0.07115\n",
            "565        20.13         28.25  ...          0.2572                  0.06637\n",
            "566        16.60         28.08  ...          0.2218                  0.07820\n",
            "567        20.60         29.33  ...          0.4087                  0.12400\n",
            "568         7.76         24.54  ...          0.2871                  0.07039\n",
            "\n",
            "[569 rows x 30 columns]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "QO2lpscHCFkD",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Adding column to data frame\n",
        "data['class'] = breast_cancer.target"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ANr8g_TOCVfz",
        "colab_type": "code",
        "outputId": "0b397d1e-a645-4913-c69a-2be309b415c1",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        }
      },
      "source": [
        "# Data is all of real numbers which needs our interpretation\n",
        "print(data.head())"
      ],
      "execution_count": 203,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "   mean radius  mean texture  ...  worst fractal dimension  class\n",
            "0        17.99         10.38  ...                  0.11890      0\n",
            "1        20.57         17.77  ...                  0.08902      0\n",
            "2        19.69         21.25  ...                  0.08758      0\n",
            "3        11.42         20.38  ...                  0.17300      0\n",
            "4        20.29         14.34  ...                  0.07678      0\n",
            "\n",
            "[5 rows x 31 columns]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Pv-96pxNCXe1",
        "colab_type": "code",
        "outputId": "d6eb7a6f-c3ca-4058-df0c-b934b1acda31",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        }
      },
      "source": [
        "# describing the dataset\n",
        "data.describe()"
      ],
      "execution_count": 204,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>mean radius</th>\n",
              "      <th>mean texture</th>\n",
              "      <th>mean perimeter</th>\n",
              "      <th>mean area</th>\n",
              "      <th>mean smoothness</th>\n",
              "      <th>mean compactness</th>\n",
              "      <th>mean concavity</th>\n",
              "      <th>mean concave points</th>\n",
              "      <th>mean symmetry</th>\n",
              "      <th>mean fractal dimension</th>\n",
              "      <th>radius error</th>\n",
              "      <th>texture error</th>\n",
              "      <th>perimeter error</th>\n",
              "      <th>area error</th>\n",
              "      <th>smoothness error</th>\n",
              "      <th>compactness error</th>\n",
              "      <th>concavity error</th>\n",
              "      <th>concave points error</th>\n",
              "      <th>symmetry error</th>\n",
              "      <th>fractal dimension error</th>\n",
              "      <th>worst radius</th>\n",
              "      <th>worst texture</th>\n",
              "      <th>worst perimeter</th>\n",
              "      <th>worst area</th>\n",
              "      <th>worst smoothness</th>\n",
              "      <th>worst compactness</th>\n",
              "      <th>worst concavity</th>\n",
              "      <th>worst concave points</th>\n",
              "      <th>worst symmetry</th>\n",
              "      <th>worst fractal dimension</th>\n",
              "      <th>class</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>count</th>\n",
              "      <td>569.000000</td>\n",
              "      <td>569.000000</td>\n",
              "      <td>569.000000</td>\n",
              "      <td>569.000000</td>\n",
              "      <td>569.000000</td>\n",
              "      <td>569.000000</td>\n",
              "      <td>569.000000</td>\n",
              "      <td>569.000000</td>\n",
              "      <td>569.000000</td>\n",
              "      <td>569.000000</td>\n",
              "      <td>569.000000</td>\n",
              "      <td>569.000000</td>\n",
              "      <td>569.000000</td>\n",
              "      <td>569.000000</td>\n",
              "      <td>569.000000</td>\n",
              "      <td>569.000000</td>\n",
              "      <td>569.000000</td>\n",
              "      <td>569.000000</td>\n",
              "      <td>569.000000</td>\n",
              "      <td>569.000000</td>\n",
              "      <td>569.000000</td>\n",
              "      <td>569.000000</td>\n",
              "      <td>569.000000</td>\n",
              "      <td>569.000000</td>\n",
              "      <td>569.000000</td>\n",
              "      <td>569.000000</td>\n",
              "      <td>569.000000</td>\n",
              "      <td>569.000000</td>\n",
              "      <td>569.000000</td>\n",
              "      <td>569.000000</td>\n",
              "      <td>569.000000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>mean</th>\n",
              "      <td>14.127292</td>\n",
              "      <td>19.289649</td>\n",
              "      <td>91.969033</td>\n",
              "      <td>654.889104</td>\n",
              "      <td>0.096360</td>\n",
              "      <td>0.104341</td>\n",
              "      <td>0.088799</td>\n",
              "      <td>0.048919</td>\n",
              "      <td>0.181162</td>\n",
              "      <td>0.062798</td>\n",
              "      <td>0.405172</td>\n",
              "      <td>1.216853</td>\n",
              "      <td>2.866059</td>\n",
              "      <td>40.337079</td>\n",
              "      <td>0.007041</td>\n",
              "      <td>0.025478</td>\n",
              "      <td>0.031894</td>\n",
              "      <td>0.011796</td>\n",
              "      <td>0.020542</td>\n",
              "      <td>0.003795</td>\n",
              "      <td>16.269190</td>\n",
              "      <td>25.677223</td>\n",
              "      <td>107.261213</td>\n",
              "      <td>880.583128</td>\n",
              "      <td>0.132369</td>\n",
              "      <td>0.254265</td>\n",
              "      <td>0.272188</td>\n",
              "      <td>0.114606</td>\n",
              "      <td>0.290076</td>\n",
              "      <td>0.083946</td>\n",
              "      <td>0.627417</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>std</th>\n",
              "      <td>3.524049</td>\n",
              "      <td>4.301036</td>\n",
              "      <td>24.298981</td>\n",
              "      <td>351.914129</td>\n",
              "      <td>0.014064</td>\n",
              "      <td>0.052813</td>\n",
              "      <td>0.079720</td>\n",
              "      <td>0.038803</td>\n",
              "      <td>0.027414</td>\n",
              "      <td>0.007060</td>\n",
              "      <td>0.277313</td>\n",
              "      <td>0.551648</td>\n",
              "      <td>2.021855</td>\n",
              "      <td>45.491006</td>\n",
              "      <td>0.003003</td>\n",
              "      <td>0.017908</td>\n",
              "      <td>0.030186</td>\n",
              "      <td>0.006170</td>\n",
              "      <td>0.008266</td>\n",
              "      <td>0.002646</td>\n",
              "      <td>4.833242</td>\n",
              "      <td>6.146258</td>\n",
              "      <td>33.602542</td>\n",
              "      <td>569.356993</td>\n",
              "      <td>0.022832</td>\n",
              "      <td>0.157336</td>\n",
              "      <td>0.208624</td>\n",
              "      <td>0.065732</td>\n",
              "      <td>0.061867</td>\n",
              "      <td>0.018061</td>\n",
              "      <td>0.483918</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>min</th>\n",
              "      <td>6.981000</td>\n",
              "      <td>9.710000</td>\n",
              "      <td>43.790000</td>\n",
              "      <td>143.500000</td>\n",
              "      <td>0.052630</td>\n",
              "      <td>0.019380</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.106000</td>\n",
              "      <td>0.049960</td>\n",
              "      <td>0.111500</td>\n",
              "      <td>0.360200</td>\n",
              "      <td>0.757000</td>\n",
              "      <td>6.802000</td>\n",
              "      <td>0.001713</td>\n",
              "      <td>0.002252</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.007882</td>\n",
              "      <td>0.000895</td>\n",
              "      <td>7.930000</td>\n",
              "      <td>12.020000</td>\n",
              "      <td>50.410000</td>\n",
              "      <td>185.200000</td>\n",
              "      <td>0.071170</td>\n",
              "      <td>0.027290</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.156500</td>\n",
              "      <td>0.055040</td>\n",
              "      <td>0.000000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>25%</th>\n",
              "      <td>11.700000</td>\n",
              "      <td>16.170000</td>\n",
              "      <td>75.170000</td>\n",
              "      <td>420.300000</td>\n",
              "      <td>0.086370</td>\n",
              "      <td>0.064920</td>\n",
              "      <td>0.029560</td>\n",
              "      <td>0.020310</td>\n",
              "      <td>0.161900</td>\n",
              "      <td>0.057700</td>\n",
              "      <td>0.232400</td>\n",
              "      <td>0.833900</td>\n",
              "      <td>1.606000</td>\n",
              "      <td>17.850000</td>\n",
              "      <td>0.005169</td>\n",
              "      <td>0.013080</td>\n",
              "      <td>0.015090</td>\n",
              "      <td>0.007638</td>\n",
              "      <td>0.015160</td>\n",
              "      <td>0.002248</td>\n",
              "      <td>13.010000</td>\n",
              "      <td>21.080000</td>\n",
              "      <td>84.110000</td>\n",
              "      <td>515.300000</td>\n",
              "      <td>0.116600</td>\n",
              "      <td>0.147200</td>\n",
              "      <td>0.114500</td>\n",
              "      <td>0.064930</td>\n",
              "      <td>0.250400</td>\n",
              "      <td>0.071460</td>\n",
              "      <td>0.000000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>50%</th>\n",
              "      <td>13.370000</td>\n",
              "      <td>18.840000</td>\n",
              "      <td>86.240000</td>\n",
              "      <td>551.100000</td>\n",
              "      <td>0.095870</td>\n",
              "      <td>0.092630</td>\n",
              "      <td>0.061540</td>\n",
              "      <td>0.033500</td>\n",
              "      <td>0.179200</td>\n",
              "      <td>0.061540</td>\n",
              "      <td>0.324200</td>\n",
              "      <td>1.108000</td>\n",
              "      <td>2.287000</td>\n",
              "      <td>24.530000</td>\n",
              "      <td>0.006380</td>\n",
              "      <td>0.020450</td>\n",
              "      <td>0.025890</td>\n",
              "      <td>0.010930</td>\n",
              "      <td>0.018730</td>\n",
              "      <td>0.003187</td>\n",
              "      <td>14.970000</td>\n",
              "      <td>25.410000</td>\n",
              "      <td>97.660000</td>\n",
              "      <td>686.500000</td>\n",
              "      <td>0.131300</td>\n",
              "      <td>0.211900</td>\n",
              "      <td>0.226700</td>\n",
              "      <td>0.099930</td>\n",
              "      <td>0.282200</td>\n",
              "      <td>0.080040</td>\n",
              "      <td>1.000000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>75%</th>\n",
              "      <td>15.780000</td>\n",
              "      <td>21.800000</td>\n",
              "      <td>104.100000</td>\n",
              "      <td>782.700000</td>\n",
              "      <td>0.105300</td>\n",
              "      <td>0.130400</td>\n",
              "      <td>0.130700</td>\n",
              "      <td>0.074000</td>\n",
              "      <td>0.195700</td>\n",
              "      <td>0.066120</td>\n",
              "      <td>0.478900</td>\n",
              "      <td>1.474000</td>\n",
              "      <td>3.357000</td>\n",
              "      <td>45.190000</td>\n",
              "      <td>0.008146</td>\n",
              "      <td>0.032450</td>\n",
              "      <td>0.042050</td>\n",
              "      <td>0.014710</td>\n",
              "      <td>0.023480</td>\n",
              "      <td>0.004558</td>\n",
              "      <td>18.790000</td>\n",
              "      <td>29.720000</td>\n",
              "      <td>125.400000</td>\n",
              "      <td>1084.000000</td>\n",
              "      <td>0.146000</td>\n",
              "      <td>0.339100</td>\n",
              "      <td>0.382900</td>\n",
              "      <td>0.161400</td>\n",
              "      <td>0.317900</td>\n",
              "      <td>0.092080</td>\n",
              "      <td>1.000000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>max</th>\n",
              "      <td>28.110000</td>\n",
              "      <td>39.280000</td>\n",
              "      <td>188.500000</td>\n",
              "      <td>2501.000000</td>\n",
              "      <td>0.163400</td>\n",
              "      <td>0.345400</td>\n",
              "      <td>0.426800</td>\n",
              "      <td>0.201200</td>\n",
              "      <td>0.304000</td>\n",
              "      <td>0.097440</td>\n",
              "      <td>2.873000</td>\n",
              "      <td>4.885000</td>\n",
              "      <td>21.980000</td>\n",
              "      <td>542.200000</td>\n",
              "      <td>0.031130</td>\n",
              "      <td>0.135400</td>\n",
              "      <td>0.396000</td>\n",
              "      <td>0.052790</td>\n",
              "      <td>0.078950</td>\n",
              "      <td>0.029840</td>\n",
              "      <td>36.040000</td>\n",
              "      <td>49.540000</td>\n",
              "      <td>251.200000</td>\n",
              "      <td>4254.000000</td>\n",
              "      <td>0.222600</td>\n",
              "      <td>1.058000</td>\n",
              "      <td>1.252000</td>\n",
              "      <td>0.291000</td>\n",
              "      <td>0.663800</td>\n",
              "      <td>0.207500</td>\n",
              "      <td>1.000000</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "       mean radius  mean texture  ...  worst fractal dimension       class\n",
              "count   569.000000    569.000000  ...               569.000000  569.000000\n",
              "mean     14.127292     19.289649  ...                 0.083946    0.627417\n",
              "std       3.524049      4.301036  ...                 0.018061    0.483918\n",
              "min       6.981000      9.710000  ...                 0.055040    0.000000\n",
              "25%      11.700000     16.170000  ...                 0.071460    0.000000\n",
              "50%      13.370000     18.840000  ...                 0.080040    1.000000\n",
              "75%      15.780000     21.800000  ...                 0.092080    1.000000\n",
              "max      28.110000     39.280000  ...                 0.207500    1.000000\n",
              "\n",
              "[8 rows x 31 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 204
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_gZSkAAxCpFe",
        "colab_type": "code",
        "outputId": "eb492c0e-9122-421c-868d-89d3eed58da4",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        }
      },
      "source": [
        "# the number of rows with label 1 is 357 (malignant)\n",
        "# the number of rows with label 0 is 212 (benign)\n",
        "# We can see the mean of the class is 0.627417 which shows us that around 63% of\n",
        "# data belongs to malignant class or 1\n",
        "print(data['class'].value_counts())"
      ],
      "execution_count": 205,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "1    357\n",
            "0    212\n",
            "Name: class, dtype: int64\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "iWrUgu7aC0ym",
        "colab_type": "code",
        "outputId": "b6d84dee-dd52-4262-f1b3-261b8ec64bd1",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        }
      },
      "source": [
        "# 357 malignant data and 212 benign data\n",
        "print(breast_cancer.target_names[0])\n",
        "print(breast_cancer.target_names[1])\n",
        "#print(breast_cancer.target_names)"
      ],
      "execution_count": 206,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "malignant\n",
            "benign\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qqbSwlNSDD0_",
        "colab_type": "code",
        "outputId": "e3882fb3-ee72-4c7f-945d-34cecbd520f6",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        }
      },
      "source": [
        "# Grouping the classes to compute means across 30 features of breast cancer data\n",
        "data.groupby('class').mean()"
      ],
      "execution_count": 207,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>mean radius</th>\n",
              "      <th>mean texture</th>\n",
              "      <th>mean perimeter</th>\n",
              "      <th>mean area</th>\n",
              "      <th>mean smoothness</th>\n",
              "      <th>mean compactness</th>\n",
              "      <th>mean concavity</th>\n",
              "      <th>mean concave points</th>\n",
              "      <th>mean symmetry</th>\n",
              "      <th>mean fractal dimension</th>\n",
              "      <th>radius error</th>\n",
              "      <th>texture error</th>\n",
              "      <th>perimeter error</th>\n",
              "      <th>area error</th>\n",
              "      <th>smoothness error</th>\n",
              "      <th>compactness error</th>\n",
              "      <th>concavity error</th>\n",
              "      <th>concave points error</th>\n",
              "      <th>symmetry error</th>\n",
              "      <th>fractal dimension error</th>\n",
              "      <th>worst radius</th>\n",
              "      <th>worst texture</th>\n",
              "      <th>worst perimeter</th>\n",
              "      <th>worst area</th>\n",
              "      <th>worst smoothness</th>\n",
              "      <th>worst compactness</th>\n",
              "      <th>worst concavity</th>\n",
              "      <th>worst concave points</th>\n",
              "      <th>worst symmetry</th>\n",
              "      <th>worst fractal dimension</th>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>class</th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>17.462830</td>\n",
              "      <td>21.604906</td>\n",
              "      <td>115.365377</td>\n",
              "      <td>978.376415</td>\n",
              "      <td>0.102898</td>\n",
              "      <td>0.145188</td>\n",
              "      <td>0.160775</td>\n",
              "      <td>0.087990</td>\n",
              "      <td>0.192909</td>\n",
              "      <td>0.062680</td>\n",
              "      <td>0.609083</td>\n",
              "      <td>1.210915</td>\n",
              "      <td>4.323929</td>\n",
              "      <td>72.672406</td>\n",
              "      <td>0.006780</td>\n",
              "      <td>0.032281</td>\n",
              "      <td>0.041824</td>\n",
              "      <td>0.015060</td>\n",
              "      <td>0.020472</td>\n",
              "      <td>0.004062</td>\n",
              "      <td>21.134811</td>\n",
              "      <td>29.318208</td>\n",
              "      <td>141.370330</td>\n",
              "      <td>1422.286321</td>\n",
              "      <td>0.144845</td>\n",
              "      <td>0.374824</td>\n",
              "      <td>0.450606</td>\n",
              "      <td>0.182237</td>\n",
              "      <td>0.323468</td>\n",
              "      <td>0.091530</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>12.146524</td>\n",
              "      <td>17.914762</td>\n",
              "      <td>78.075406</td>\n",
              "      <td>462.790196</td>\n",
              "      <td>0.092478</td>\n",
              "      <td>0.080085</td>\n",
              "      <td>0.046058</td>\n",
              "      <td>0.025717</td>\n",
              "      <td>0.174186</td>\n",
              "      <td>0.062867</td>\n",
              "      <td>0.284082</td>\n",
              "      <td>1.220380</td>\n",
              "      <td>2.000321</td>\n",
              "      <td>21.135148</td>\n",
              "      <td>0.007196</td>\n",
              "      <td>0.021438</td>\n",
              "      <td>0.025997</td>\n",
              "      <td>0.009858</td>\n",
              "      <td>0.020584</td>\n",
              "      <td>0.003636</td>\n",
              "      <td>13.379801</td>\n",
              "      <td>23.515070</td>\n",
              "      <td>87.005938</td>\n",
              "      <td>558.899440</td>\n",
              "      <td>0.124959</td>\n",
              "      <td>0.182673</td>\n",
              "      <td>0.166238</td>\n",
              "      <td>0.074444</td>\n",
              "      <td>0.270246</td>\n",
              "      <td>0.079442</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "       mean radius  mean texture  ...  worst symmetry  worst fractal dimension\n",
              "class                             ...                                         \n",
              "0        17.462830     21.604906  ...        0.323468                 0.091530\n",
              "1        12.146524     17.914762  ...        0.270246                 0.079442\n",
              "\n",
              "[2 rows x 30 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 207
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "MfmGSVAdFMV5",
        "colab_type": "text"
      },
      "source": [
        "# Train and test split"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "x38p_3T2D02K",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Model generalizes when it sees test data from what it has learned\n",
        "# (should not overfit or underfit  from what the model has previously learned)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "RQXz0kGzGEd_",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# from model selection package import train test split\n",
        "from sklearn.model_selection import train_test_split"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "5zVcHR7RGcC8",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Dropping the class column from our dataframe data\n",
        "# since it is column we say axis = 1 and if it is row then axis = 0\n",
        "X = data.drop('class', axis = 1)\n",
        "Y = data['class']"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "CR-xJXQxHBdN",
        "colab_type": "code",
        "outputId": "df06e730-31ec-42aa-9a3b-7c804c1518a8",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        }
      },
      "source": [
        "# we already had X and Y as numpy\n",
        "# but now we need them as dataframe just because we can do operation conveniently\n",
        "print(type(X))\n",
        "print(type(Y))"
      ],
      "execution_count": 211,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "<class 'pandas.core.frame.DataFrame'>\n",
            "<class 'pandas.core.series.Series'>\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "HfJJ4tYIHgTA",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Now we have to split the data into test and train\n",
        "X_train, X_test, Y_train, Y_test = train_test_split(X, Y)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "CBzjjOnLH_YD",
        "colab_type": "code",
        "outputId": "b78f9ea9-1f14-480b-9e97-83c1f51bca19",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        }
      },
      "source": [
        "# roughly 80 percent goes for training and 20 percent goes to testing\n",
        "print(X.shape, X_train.shape, X_test.shape)\n",
        "print(Y.shape, Y_train.shape, Y_test.shape)"
      ],
      "execution_count": 213,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(569, 30) (426, 30) (143, 30)\n",
            "(569,) (426,) (143,)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Nhwq2veRILs6",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# if we want to change the test size of the data\n",
        "X_train, X_test, Y_train, Y_test = train_test_split(X, Y, test_size = 0.1)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "JEpVrV_II7cl",
        "colab_type": "code",
        "outputId": "2a6ab5b8-7cea-454b-c3e5-29e8ff60facc",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        }
      },
      "source": [
        "print(X.shape, X_train.shape, X_test.shape)\n",
        "print(Y.shape, Y_train.shape, Y_test.shape)"
      ],
      "execution_count": 215,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(569, 30) (512, 30) (57, 30)\n",
            "(569,) (512,) (57,)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_PWfY8CgI9td",
        "colab_type": "code",
        "outputId": "7cec23ab-4c9f-4c65-e600-4dbfc48829a2",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        }
      },
      "source": [
        "# We know already from the dataset that there are more malignant data than benign\n",
        "# data\n",
        "\n",
        "# What we can infer here is that  the mean of malignant in test data is less when\n",
        "# compared to original data and training data and therefore we need to do \n",
        "# something here in the test data that resembles the mean of original data\n",
        "print(Y.mean(), Y_train.mean(), Y_test.mean())"
      ],
      "execution_count": 216,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "0.6274165202108963 0.62890625 0.6140350877192983\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "MEP9Im8pJMzr",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# To do it we stratify the data to maintain the ratio \n",
        "X_train, X_test, Y_train, Y_test = train_test_split(X, Y, test_size = 0.1, stratify = Y)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "nz-sVphjLWVa",
        "colab_type": "code",
        "outputId": "bb487fea-dc41-43fd-a798-168b1514f318",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        }
      },
      "source": [
        "# lets see the ratio of the mean of malignant is at least similar or equivalent across\n",
        "# I think the data is already stratified\n",
        "print(Y.mean(), Y_train.mean(), Y_test.mean())"
      ],
      "execution_count": 218,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "0.6274165202108963 0.626953125 0.631578947368421\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "A0mFwarVLkUU",
        "colab_type": "code",
        "outputId": "732b08c2-7b47-4755-eb5c-c3b3473c3e61",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        }
      },
      "source": [
        "# Let us see the mean in the feature part of the breast cancer data\n",
        "# So it seems the data that we have divided right now does not seem a lot\n",
        "# That is variation does not seem a lot\n",
        "print(X.mean(), X_train.mean(), X_test.mean())"
      ],
      "execution_count": 219,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "mean radius                 14.127292\n",
            "mean texture                19.289649\n",
            "mean perimeter              91.969033\n",
            "mean area                  654.889104\n",
            "mean smoothness              0.096360\n",
            "mean compactness             0.104341\n",
            "mean concavity               0.088799\n",
            "mean concave points          0.048919\n",
            "mean symmetry                0.181162\n",
            "mean fractal dimension       0.062798\n",
            "radius error                 0.405172\n",
            "texture error                1.216853\n",
            "perimeter error              2.866059\n",
            "area error                  40.337079\n",
            "smoothness error             0.007041\n",
            "compactness error            0.025478\n",
            "concavity error              0.031894\n",
            "concave points error         0.011796\n",
            "symmetry error               0.020542\n",
            "fractal dimension error      0.003795\n",
            "worst radius                16.269190\n",
            "worst texture               25.677223\n",
            "worst perimeter            107.261213\n",
            "worst area                 880.583128\n",
            "worst smoothness             0.132369\n",
            "worst compactness            0.254265\n",
            "worst concavity              0.272188\n",
            "worst concave points         0.114606\n",
            "worst symmetry               0.290076\n",
            "worst fractal dimension      0.083946\n",
            "dtype: float64 mean radius                 14.078154\n",
            "mean texture                19.304492\n",
            "mean perimeter              91.640293\n",
            "mean area                  649.765820\n",
            "mean smoothness              0.096239\n",
            "mean compactness             0.104154\n",
            "mean concavity               0.088661\n",
            "mean concave points          0.048588\n",
            "mean symmetry                0.180927\n",
            "mean fractal dimension       0.062870\n",
            "radius error                 0.402142\n",
            "texture error                1.230231\n",
            "perimeter error              2.841138\n",
            "area error                  39.670604\n",
            "smoothness error             0.007004\n",
            "compactness error            0.025615\n",
            "concavity error              0.032089\n",
            "concave points error         0.011736\n",
            "symmetry error               0.020512\n",
            "fractal dimension error      0.003802\n",
            "worst radius                16.241463\n",
            "worst texture               25.795508\n",
            "worst perimeter            107.084727\n",
            "worst area                 877.703711\n",
            "worst smoothness             0.132308\n",
            "worst compactness            0.255809\n",
            "worst concavity              0.273147\n",
            "worst concave points         0.114282\n",
            "worst symmetry               0.290801\n",
            "worst fractal dimension      0.084255\n",
            "dtype: float64 mean radius                 14.568667\n",
            "mean texture                19.156316\n",
            "mean perimeter              94.921930\n",
            "mean area                  700.908772\n",
            "mean smoothness              0.097447\n",
            "mean compactness             0.106023\n",
            "mean concavity               0.090040\n",
            "mean concave points          0.051890\n",
            "mean symmetry                0.183270\n",
            "mean fractal dimension       0.062143\n",
            "radius error                 0.432389\n",
            "texture error                1.096688\n",
            "perimeter error              3.089912\n",
            "area error                  46.323667\n",
            "smoothness error             0.007376\n",
            "compactness error            0.024252\n",
            "concavity error              0.030137\n",
            "concave points error         0.012332\n",
            "symmetry error               0.020813\n",
            "fractal dimension error      0.003733\n",
            "worst radius                16.518246\n",
            "worst texture               24.614737\n",
            "worst perimeter            108.846491\n",
            "worst area                 906.447368\n",
            "worst smoothness             0.132914\n",
            "worst compactness            0.240393\n",
            "worst concavity              0.263580\n",
            "worst concave points         0.117515\n",
            "worst symmetry               0.283561\n",
            "worst fractal dimension      0.081168\n",
            "dtype: float64\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_VtguskYMOHU",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Let us run this again\n",
        "X_train, X_test, Y_train, Y_test = train_test_split(X, Y, test_size = 0.1, stratify = Y)\n",
        "#print(Y_test)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Mip8lfZvM64I",
        "colab_type": "code",
        "outputId": "c0aeb6cd-f130-4fff-cfac-8e4a78bdf18e",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        }
      },
      "source": [
        "# Again run this\n",
        "print(X.mean(), X_train.mean(), X_test.mean())"
      ],
      "execution_count": 221,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "mean radius                 14.127292\n",
            "mean texture                19.289649\n",
            "mean perimeter              91.969033\n",
            "mean area                  654.889104\n",
            "mean smoothness              0.096360\n",
            "mean compactness             0.104341\n",
            "mean concavity               0.088799\n",
            "mean concave points          0.048919\n",
            "mean symmetry                0.181162\n",
            "mean fractal dimension       0.062798\n",
            "radius error                 0.405172\n",
            "texture error                1.216853\n",
            "perimeter error              2.866059\n",
            "area error                  40.337079\n",
            "smoothness error             0.007041\n",
            "compactness error            0.025478\n",
            "concavity error              0.031894\n",
            "concave points error         0.011796\n",
            "symmetry error               0.020542\n",
            "fractal dimension error      0.003795\n",
            "worst radius                16.269190\n",
            "worst texture               25.677223\n",
            "worst perimeter            107.261213\n",
            "worst area                 880.583128\n",
            "worst smoothness             0.132369\n",
            "worst compactness            0.254265\n",
            "worst concavity              0.272188\n",
            "worst concave points         0.114606\n",
            "worst symmetry               0.290076\n",
            "worst fractal dimension      0.083946\n",
            "dtype: float64 mean radius                 14.133420\n",
            "mean texture                19.290937\n",
            "mean perimeter              92.002012\n",
            "mean area                  656.368359\n",
            "mean smoothness              0.096396\n",
            "mean compactness             0.104164\n",
            "mean concavity               0.089145\n",
            "mean concave points          0.049137\n",
            "mean symmetry                0.181002\n",
            "mean fractal dimension       0.062733\n",
            "radius error                 0.405284\n",
            "texture error                1.216357\n",
            "perimeter error              2.864253\n",
            "area error                  40.477035\n",
            "smoothness error             0.007041\n",
            "compactness error            0.025330\n",
            "concavity error              0.031696\n",
            "concave points error         0.011717\n",
            "symmetry error               0.020433\n",
            "fractal dimension error      0.003748\n",
            "worst radius                16.280373\n",
            "worst texture               25.686289\n",
            "worst perimeter            107.333594\n",
            "worst area                 883.094922\n",
            "worst smoothness             0.132378\n",
            "worst compactness            0.254846\n",
            "worst concavity              0.273257\n",
            "worst concave points         0.114791\n",
            "worst symmetry               0.290050\n",
            "worst fractal dimension      0.083846\n",
            "dtype: float64 mean radius                 14.072246\n",
            "mean texture                19.278070\n",
            "mean perimeter              91.672807\n",
            "mean area                  641.601754\n",
            "mean smoothness              0.096036\n",
            "mean compactness             0.105929\n",
            "mean concavity               0.085696\n",
            "mean concave points          0.046962\n",
            "mean symmetry                0.182596\n",
            "mean fractal dimension       0.063376\n",
            "radius error                 0.404170\n",
            "texture error                1.221311\n",
            "perimeter error              2.882284\n",
            "area error                  39.079930\n",
            "smoothness error             0.007042\n",
            "compactness error            0.026806\n",
            "concavity error              0.033673\n",
            "concave points error         0.012510\n",
            "symmetry error               0.021523\n",
            "fractal dimension error      0.004213\n",
            "worst radius                16.168737\n",
            "worst texture               25.595789\n",
            "worst perimeter            106.611053\n",
            "worst area                 858.021053\n",
            "worst smoothness             0.132287\n",
            "worst compactness            0.249048\n",
            "worst concavity              0.262595\n",
            "worst concave points         0.112943\n",
            "worst symmetry               0.290304\n",
            "worst fractal dimension      0.084843\n",
            "dtype: float64\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "XEsQ3-fdNCIT",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# What we can see is everytime we run the stratifying command the mean ratio \n",
        "# varies in some fashion which is not advised because in machine learning we\n",
        "# want the results to be reproducible\n",
        "\n",
        "''' Very important '''\n",
        "# If we run the model and get some good accuracy and if we re-run the model\n",
        "# and do not get the same accuracy again then we are in a great trouble\n",
        "# This leads to not put the code in the production\n",
        "\n",
        "# We use random_state and use it as a seed (what it does is the split that happens\n",
        "# is always deterministic (no matter how many times we run this it will always split the \n",
        "# in the same ))\n",
        "# This ensures there is no randomness involved in the split\n",
        "X_train, X_test, Y_train, Y_test = train_test_split(X, Y, test_size = 0.1, stratify = Y, random_state = 1)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Y0rcr2-lO4M-",
        "colab_type": "code",
        "outputId": "808cf818-15a1-44bd-d764-803565e9566b",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        }
      },
      "source": [
        "# Let us see it\n",
        "# So we can see how many times we run this the split will be in the same ratio\n",
        "# This leads us to deploy the code in production\n",
        "print(X.mean(), X_train.mean(), X_test.mean())"
      ],
      "execution_count": 223,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "mean radius                 14.127292\n",
            "mean texture                19.289649\n",
            "mean perimeter              91.969033\n",
            "mean area                  654.889104\n",
            "mean smoothness              0.096360\n",
            "mean compactness             0.104341\n",
            "mean concavity               0.088799\n",
            "mean concave points          0.048919\n",
            "mean symmetry                0.181162\n",
            "mean fractal dimension       0.062798\n",
            "radius error                 0.405172\n",
            "texture error                1.216853\n",
            "perimeter error              2.866059\n",
            "area error                  40.337079\n",
            "smoothness error             0.007041\n",
            "compactness error            0.025478\n",
            "concavity error              0.031894\n",
            "concave points error         0.011796\n",
            "symmetry error               0.020542\n",
            "fractal dimension error      0.003795\n",
            "worst radius                16.269190\n",
            "worst texture               25.677223\n",
            "worst perimeter            107.261213\n",
            "worst area                 880.583128\n",
            "worst smoothness             0.132369\n",
            "worst compactness            0.254265\n",
            "worst concavity              0.272188\n",
            "worst concave points         0.114606\n",
            "worst symmetry               0.290076\n",
            "worst fractal dimension      0.083946\n",
            "dtype: float64 mean radius                 14.058656\n",
            "mean texture                19.309668\n",
            "mean perimeter              91.530488\n",
            "mean area                  648.097266\n",
            "mean smoothness              0.096568\n",
            "mean compactness             0.105144\n",
            "mean concavity               0.089342\n",
            "mean concave points          0.048892\n",
            "mean symmetry                0.181961\n",
            "mean fractal dimension       0.062979\n",
            "radius error                 0.403659\n",
            "texture error                1.206856\n",
            "perimeter error              2.861173\n",
            "area error                  39.935506\n",
            "smoothness error             0.007067\n",
            "compactness error            0.025681\n",
            "concavity error              0.032328\n",
            "concave points error         0.011963\n",
            "symmetry error               0.020584\n",
            "fractal dimension error      0.003815\n",
            "worst radius                16.194275\n",
            "worst texture               25.644902\n",
            "worst perimeter            106.757715\n",
            "worst area                 871.647852\n",
            "worst smoothness             0.132592\n",
            "worst compactness            0.257415\n",
            "worst concavity              0.275623\n",
            "worst concave points         0.115454\n",
            "worst symmetry               0.291562\n",
            "worst fractal dimension      0.084402\n",
            "dtype: float64 mean radius                 14.743807\n",
            "mean texture                19.109825\n",
            "mean perimeter              95.908246\n",
            "mean area                  715.896491\n",
            "mean smoothness              0.094496\n",
            "mean compactness             0.097130\n",
            "mean concavity               0.083923\n",
            "mean concave points          0.049159\n",
            "mean symmetry                0.173981\n",
            "mean fractal dimension       0.061169\n",
            "radius error                 0.418767\n",
            "texture error                1.306656\n",
            "perimeter error              2.909946\n",
            "area error                  43.944193\n",
            "smoothness error             0.006809\n",
            "compactness error            0.023659\n",
            "concavity error              0.027989\n",
            "concave points error         0.010293\n",
            "symmetry error               0.020169\n",
            "fractal dimension error      0.003618\n",
            "worst radius                16.942105\n",
            "worst texture               25.967544\n",
            "worst perimeter            111.783860\n",
            "worst area                 960.843860\n",
            "worst smoothness             0.130357\n",
            "worst compactness            0.225973\n",
            "worst concavity              0.241340\n",
            "worst concave points         0.106994\n",
            "worst symmetry               0.276719\n",
            "worst fractal dimension      0.079852\n",
            "dtype: float64\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "bZNZ86t_Qbyv",
        "colab_type": "text"
      },
      "source": [
        "# Binarization of the input"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "iAoHC31lO_YM",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# MP neuron model takes only binary inputs\n",
        "# But our breast cancer data is a continuous data set and so we need to convert \n",
        "# them to binary inputs\n",
        "\n",
        "# Let us plot the data\n",
        "import matplotlib.pyplot as plt"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ZLuMFQRqQ9MP",
        "colab_type": "code",
        "outputId": "e1678501-ae1e-4755-d642-f3b475c16a79",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        }
      },
      "source": [
        "# plot features of training data\n",
        "plt.plot(X_train, '*')\n",
        "plt.show()"
      ],
      "execution_count": 225,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ibCFARRuRMl3",
        "colab_type": "code",
        "outputId": "bb9966c4-a7a7-4f02-d13f-9b5bdbfc93cf",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        }
      },
      "source": [
        "# Also plot in this way for better interpretation\n",
        "# For each feature how the values vary can seen\n",
        "plt.plot(X_train.T, '*')\n",
        "plt.xticks(rotation = 'vertical')\n",
        "plt.show()"
      ],
      "execution_count": 226,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7S-5EYivUegf",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Trying to binarise only one column of the training data which is mean area\n",
        "# Finds a threshold a binarizes some of the value to 0 and others to 1\n",
        "X_binarise_train_mean_area = X_train['mean area'].map(lambda x: 0 if x < 1000 else 1)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3o0TSFy9Uyzm",
        "colab_type": "code",
        "outputId": "ae77289d-ff7f-4222-944d-9b193f624656",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        }
      },
      "source": [
        "plt.plot(X_binarise_train_mean_area, '*')"
      ],
      "execution_count": 228,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[<matplotlib.lines.Line2D at 0x7f320f9ac898>]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 228
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXQAAAD4CAYAAAD8Zh1EAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+j8jraAAAZJElEQVR4nO3de5Bc5Xnn8e/T3XMTI40uM1IkzQiBI9mRjBbjWS5lqizY2AhwQWS8CXJwLkVFXm9YvLVeG1F2YcG6YoyrciErEuTESQWvzRJjb7SOiGxjsqVQGBgFWyAJGRkEuqvRZaRBc9HMPPvHOT3q6enLmVGPhn7z+1RNqc857znned/z9k+t09Nqc3dERKT2paa6ABERqQ4FuohIIBToIiKBUKCLiARCgS4iEojMVJ24tbXVFy9ePFWnFxGpSdu2bXvb3duKbZuyQF+8eDFdXV1TdXoRkZpkZm+W2qZbLiIigVCgi4gEQoEuIhIIBbqISCAU6CIigagY6Gb2TTM7amavlNhuZvawme0xs+1mdkX1yzzn6Kk+bn54K8vu+yc+9udb2Xmwm9989DmOnu7j6Km+oo/z9/2NDc+y+pFnR+2Xv331hmf5jQ3PJlpfWFf+8XYe7OayL2/hX17LsnrDs9z88FY+9udbWf3IuWMUq7HU8cqNR65Ppfqd1M6D3Sz/cjSu5fbP9W3noe6S5yt3LQrHs7APpcbiX17Ljpy32Dly41zsOEnGuticKKXSOOcfc7xzZzzbc9fshj/9f6x+5NkxY1SptonOlUrHLnfMUtd7onO33BzMjX25vo48V/dkJ3wdktSYy62bHy7//Dof6fXr15dtcP/9958AvgmsXr9+/SNFtt8E3AhcDbwE/M/169f/VaUTb9y4cf3atWvHXfBXN+/i6VePcnbIOXq6nxfeOM4rB7vp7R/i+TeO8087Do95fP2vzRvZ94c7j3C4u2/UfmO2n+qjd6Dy+sK68s/3yW88z6FTffx41xHePH6G7Ol+jp7u53D3uWMU7lPueOXGI9en3oHi/U7qk994nsOnojpL9TPX7tCpaAwPnOgter78+gtrKhzP5984PqoPpcbixzuPcLp/kBfeOM7vXLN4zDl+uPMI2YIxTjKeuW3F5kS5cS83zvnH3HHo1Ljmzni2567ZsZ4BDndH8y1/jMrVPp7+JpVk3hbO2fzn2UTmbqn98udZub6OPFd3HuGNY+9M6DokqTGXW9kKz69K7r///kPr16/fWGybJfnvc81sMfADd39/kW2PAv/s7t+Jl3cDK939ULljdnZ2+nh+D/29X3qK/sHhxO0nW0Mmxe6v3FjVuhoy0T+Yih0vd76cpOct3K+Yxev+MdH+5dpVW7mxuFDHLzZ2pcZ9PPOhUtsk28czLnsfvLls7YXnnYhKfUly/vHWcz7PvSRjmPQ6lTMZY25m29y9s9i2atxDXwjsy1veH68rVshaM+sys65sNjuuk2z9wnV8eGlrye0pg7RFj9MWLQM01qVYtXweH17aOrIuX0Mm2r5yaevI/rnjrVzaysqlbWPW37B8HlvvuW6krlsuX0BjXTSU9WloqkuX7UvKYH5LIw0ZG6nx1ssXsPWe68YcL39b4Xh8dPm8MX3K73ex/YrZfPe1/EpLw5j11y1tHbX/5ruvZeHMpqLHyJ1v893Xjqo//7o0ZFLMb2ksO+nSBqvi8c2NRX2R4WyqS4+sL3ZdU0THKayn2FjnrkNOQ6b02FW6PueOObaXleZOqWMV217qmuVrn9nE5s9eO6b2wtoaMpZ4rpSSZN4Wm7Npg5XvbeOjy+dVnPNJz7n57mu5Yfm8Uc/bYn0tNZ8Lr3/S52SpGkvl1sqC51c1XNBPirr7RmAjRK/Qx7Pv3BmNtM+aVnRb2mAoPlr+37y5x63NDTgw7GP3Gxg6t30ob/uwQ/usaUXXtzU3MHd640hd0xsy9A8O05BJMTA0TGOd0Xu2dF+GHZobMhw+1TdS4/SGzMgx849XuC1/PNqaG8b0adgpu18xyxa0ML2hjsP0j1q/cNa0UfsvW9DCtCLpmn++6Fij68+1GRgaprkhQ7l/ug05tOaN7/SGDGeLvMCpSxt9g6VfaQ0THadYPYVjPTDkpCwau9ycKDV2hde78Hi57QNDw6PmJVSeO6WOVWz73OmNRa9Zvqb6NMvmt4ypPb+2lMHAkCeeK6VU6kuuTeGcHfLoLx6HinM+6TmXLWihtblh1NjD2L7Ond44Zj4bY69/kr6Vq7FUbrUXPL+qoRqBfgDoyFtuj9dV3ds9/TTFf0vObKrnyOk+DLjy0tnsP94LwKOf6uTTj3WNPP72C2+Rjd+A6JjVxIr2mXS9eZyevkFWdLTwnrbpY7YDbN9/kmxPf9n1+XX99lUX88krF/HtF97iia59LJ3XTGNdip0HT42ELMCMxjrqMym6e8+O2ieb9yZJ4fGyJd5Aebunf1RtT796hLbmhjH9TqK79ywX1ae5YtEs3jp+hqOn+8b0M9du6bxm7r5+Cfc8uZ1hd578zIdGnS+//sJr8cMdh0fV/JNXjwBw/fvmFR3f3LE2bz/IkDsXz5lG39lh9h47M+oc2Z5+WhrraKxL0zc4RH06NXKccuOZ2/Z6toe3e/qZ01w/ak6UGvdy1ye3/ZfZ02zf101zY4bOi2cnmjuljlVse+6aNdal6R8coqd/iEwKli9ooffsECfPjH1VkV/bsZ4BWpsbuLSteVxzZaLjkmtT6vmUZM4nPWf+ebrePM47/YOsaJ85pq+5+dxUl2bvsXdImXHzigXjug5JajyXW3WcjF/tFXt+na9q3EO/GbgLuAm4CnjY3a+sdMzx3kMXEZHy99ArvkI3s+8AK4FWM9sPfBmoA3D3vwQ2E4X5HuAM8PvVKVtERMajYqC7+5oK2x34w6pVJCIiE6JPioqIBEKBLiISCAW6iEggFOgiIoFQoIuIBEKBLiISCAW6iEggFOgiIoFQoIuIBEKBLiISCAW6iEggFOgiIoFQoIuIBEKBLiISCAW6iEggFOgiIoFQoIuIBEKBLiISCAW6iEggFOgiIoFQoIuIBEKBLiISCAW6iEggFOgiIoFQoIuIBEKBLiISCAW6iEggFOgiIoFQoIuIBEKBLiISCAW6iEggEgW6ma0ys91mtsfM1hXZvsjMnjGzl8xsu5ndVP1SRUSknIqBbmZpYANwI7AMWGNmywqafQl4wt0/ANwOPFLtQkVEpLwkr9CvBPa4++vuPgA8Dtxa0MaBGfHjFuBg9UoUEZEkkgT6QmBf3vL+eF2+9cAdZrYf2Az8l2IHMrO1ZtZlZl3ZbHYC5YqISCnVelN0DfC37t4O3AQ8ZmZjju3uG929090729raqnRqERGBZIF+AOjIW26P1+W7E3gCwN2fAxqB1moUKCIiySQJ9BeBJWZ2iZnVE73puamgzVvAfwAws18jCnTdUxERuYAqBrq7DwJ3AVuAXUS/zbLDzB4ws1viZp8D/sDMfg58B/g9d/fJKlpERMbKJGnk7puJ3uzMX3df3uOdwIeqW5qIiIyHPikqIhIIBbqISCAU6CIigVCgi4gEQoEuIhIIBbqISCAU6CIigVCgi4gEQoEuIhIIBbqISCAU6CIigVCgi4gEQoEuIhIIBbqISCAU6CIigVCgi4gEQoEuIhIIBbqISCAU6CIigVCgi4gEQoEuIhIIBbqISCAU6CIigVCgi4gEQoEuIhIIBbqISCAU6CIigVCgi4gEQoEuIhIIBbqISCASBbqZrTKz3Wa2x8zWlWjzm2a208x2mNm3q1umiIhUkqnUwMzSwAbgI8B+4EUz2+TuO/PaLAHuBT7k7ifMbO5kFSwiIsUleYV+JbDH3V939wHgceDWgjZ/AGxw9xMA7n60umWKiEglSQJ9IbAvb3l/vC7fUmCpmT1rZj81s1XFDmRma82sy8y6stnsxCoWEZGiqvWmaAZYAqwE1gDfMLOZhY3cfaO7d7p7Z1tbW5VOLSIikCzQDwAdecvt8bp8+4FN7n7W3d8AfkEU8CIicoEkCfQXgSVmdomZ1QO3A5sK2vwfolfnmFkr0S2Y16tYp4iIVFAx0N19ELgL2ALsAp5w9x1m9oCZ3RI32wIcM7OdwDPA59392GQVLSIiY5m7T8mJOzs7vaura0rOLSJSq8xsm7t3FtumT4qKiARCgS4iEggFuohIIBToIiKBUKCLiARCgS4iEggFuohIIBToIiKBUKCLiARCgS4iEggFuohIIBToIiKBUKCLiARCgS4iEggFuohIIBToIiKBUKCLiARCgS4iEggFuohIIBToIiKBUKCLiARCgS4iEggFuohIIBToIiKBUKCLiARCgS4iEggFuohIIBToIiKBUKCLiARCgS4iEggFuohIIBIFupmtMrPdZrbHzNaVaXebmbmZdVavRBERSaJioJtZGtgA3AgsA9aY2bIi7aYDnwWer3aRIiJSWZJX6FcCe9z9dXcfAB4Hbi3S7n8AXwP6qlifiIgklCTQFwL78pb3x+tGmNkVQIe7/2O5A5nZWjPrMrOubDY77mJFRKS0835T1MxSwB8Dn6vU1t03ununu3e2tbWd76lFRCRPkkA/AHTkLbfH63KmA+8H/tnM9gJXA5v0xqiIyIWVJNBfBJaY2SVmVg/cDmzKbXT3bndvdffF7r4Y+Clwi7t3TUrFIiJSVMVAd/dB4C5gC7ALeMLdd5jZA2Z2y2QXKCIiyWSSNHL3zcDmgnX3lWi78vzLEhGR8dInRUVEAqFAFxEJhAJdRCQQCnQRkUAo0EVEAqFAFxEJhAJdRCQQCnQRkUAo0EVEAqFAFxEJhAJdRCQQCnQRkUAo0EVEAqFAFxEJhAJdRCQQCnQRkUAo0EVEAqFAFxEJhAJdRCQQCnQRkUAo0EVEAqFAFxEJhAJdRCQQCnQRkUAo0EVEAqFAFxEJhAJdRCQQCnQRkUAo0EVEAqFAFxEJRKJAN7NVZrbbzPaY2boi2/+bme00s+1m9rSZXVz9UkVEpJyKgW5maWADcCOwDFhjZssKmr0EdLr7CuC7wEPVLlRERMpL8gr9SmCPu7/u7gPA48Ct+Q3c/Rl3PxMv/hRor26ZIiJSSZJAXwjsy1veH68r5U7gqWIbzGytmXWZWVc2m01epYiIVFTVN0XN7A6gE/h6se3uvtHdO929s62trZqnFhH5Ny+ToM0BoCNvuT1eN4qZ/TrwReDD7t5fnfJERCSpJK/QXwSWmNklZlYP3A5sym9gZh8AHgVucfej1S9TREQqqRjo7j4I3AVsAXYBT7j7DjN7wMxuiZt9HWgG/t7MfmZmm0ocTkREJkmSWy64+2Zgc8G6+/Ie/3qV6xIRkXHSJ0VFRAKhQBcRCYQCXUQkEAp0EZFAKNBFRAKhQBcRCYQCXUQkEAp0EZFAKNBFRAKhQBcRCYQCXUQkEAp0EZFAKNBFRAKhQBcRCYQCXUQkEAp0EZFAKNBFRAKhQBcRCYQCXUQkEAp0EZFAKNBFRAKhQBcRCYQCXUQkEAp0EZFAKNBFRAKhQBcRCYQCXUQkEAp0EZFAKNBFRAKhQBcRCYQCXUQkEJkkjcxsFfBnQBr4K3d/sGB7A/B3wAeBY8Bvufve6pY62tFTfXz6sW0MDA0zODTMW8fPADC/pZFD3X0AXDznItydfSd6mT+jgcOn+3n0jg/y4FOv8sbb7/DgbZdxz5MvMzQ0RCqVYtGcadSlUgwMDbP/RC+/MqOBQ919OM7iOc24O28eewezFPNb4m0O4JgZC2Y2cvBkdO73zG1m3ar38enHttExexrplGGAA0PDzt5jPRjGgplNHDzZy7APY6TAnPaZ0zhwspcFM5s4cPIMeLynwcKZTRzs7mPR7GngsO9EL5e2XRSd61vbaJ/VRF06xeCQs/ftHjDDfZh0Ks1Dt13G57+7nahkxz2qO76GvGduMw/dtoIvfv8VzgwM8tbxMyNtcn0zM777mWvA4RN/+RzuzqI50zCMN4+9M1JzylI8FI9vx+xp0dgdfwfDRsYZ4OzQMPtO9NIxqwkgPmfBmHb3sbCliUPdfTx422V84cnt4ERjkNsnvkaZlOGAxcfef7KXr338Mu793iv80cffzz1Pvoy70zF7GvXp1Mj522c1UZ+OahpzjVqiGnJjsWj2NJrqM/z3jy7l09/axrzpuXkSnTd37IGhYfadOMPFs6N5mKvl809ux4ejMzlOytJ0zG6iPpMCPzcm82c0cDCey7kxOxvP9Vwtl7Y18/VPrGD9/93J3df/Kp/51r/yF3dcwVefepXXsz2jngdvHT8zMt8WzmwamWMHT/aOHD+TTvFHq9/P57+7ndezPZhZNG/imp1hUpaO538/znB0bU71jzzHvvbxaJ6BnXueAO7DmKUwomuXSacYGo7mxYIZjRw81UfHrGmkzEatW9gS1To8PDzqeToyFpx7zlw85yIyKYu2nTjDolnTcGDfiTMsbImeO7lx9WiiYSnjodtWsO57L7NwZtPItdsfzwsz2He8l0taL+LeG9/H2se2MTQ8hFmKaLLaufna0siB7t74OXtuvC5tu4i/+f1/Dw5rH9uGGTz6qQ8yd3pjtSIx6kuuUyUbmKWBXwAfAfYDLwJr3H1nXpv/DKxw9/9kZrcDq939t8odt7Oz07u6uiZc+Je+/zLfev6tce83ozHDqb5BAOrSxtmh8v0/H/nnmmxJzpWkv0vmNvPa0Z6KbYCK7SZjfCdyzNw+1a5nItd3MsZkydxm9mR7mN4Q1XO+865wDoy35sl+Xk2GpDWfz9jecdUigJHcuuOqRXxl9WXjPo6ZbXP3zqLbEgT6NcB6d78hXr4XwN2/mtdmS9zmOTPLAIeBNi9z8IkG+nu/9BT9g8Pj3k9E5N2oIZNi91duTNy+XKAnuYe+ENiXt7w/Xle0jbsPAt3AnCKFrDWzLjPrymazSWofY+sXruOG5fNI2YR2FxF5V0gbrFo+j633XFe1Y17QN0XdfaO7d7p7Z1tb24SOMXdGI63NDQzX1r/oRERGGXJobW6o6n30JIF+AOjIW26P1xVtE99yaSF6c3RSvN3TT8esJuZNbxjXK/Vc05RF98wmS/6g1qeN+L22SflXRUPm3EHr8s51PsqVmX++hoxV7FOSLk9kWMp1M1dT2qJj534KtxdbHu81qtTcSjyuprRBXcpoacowozEzap6PRwpoyHteTKtLV63GqTCRa5kuMRfqzzMvUvHxmupS3HzZfG6+bD4ds5rI9vSf13ELJfktlxeBJWZ2CVFw3w58sqDNJuB3geeATwA/KXf//Hw9+qmit49ERP5Nqxjo7j5oZncBW4h+bfGb7r7DzB4Autx9E/DXwGNmtgc4ThT6IiJyASX6PXR33wxsLlh3X97jPuA/Vrc0EREZD31SVEQkEAp0EZFAKNBFRAKhQBcRCUTFj/5P2onNssCbE9y9FXi7iuW8W4TYrxD7BOpXLQmtTxe7e9FPZk5ZoJ8PM+sq9X8Z1LIQ+xVin0D9qiUh9qkU3XIREQmEAl1EJBC1Gugbp7qASRJiv0LsE6hftSTEPhVVk/fQRURkrFp9hS4iIgUU6CIigai5QDezVWa228z2mNm6qa5nPMzsm2Z21MxeyVs328x+ZGavxX/OitebmT0c93O7mV0xdZWXZmYdZvaMme00sx1m9tl4fc32y8wazewFM/t53Kf74/WXmNnzce3/28zq4/UN8fKeePviqay/EjNLm9lLZvaDeLnm+2Vme83sZTP7mZl1xetqdg5OVE0FevyF1RuAG4FlwBozWza1VY3L3wKrCtatA5529yXA0/EyRH1cEv+sBf7iAtU4XoPA59x9GXA18IfxNanlfvUD17v7vwMuB1aZ2dXA14A/cfdfBU4Ad8bt7wROxOv/JG73bvZZYFfecij9us7dL8/7nfNanoMT4+418wNcA2zJW74XuHeq6xpnHxYDr+Qt7wbmx4/nA7vjx48Ca4q1ezf/AP8AfCSUfgHTgH8FriL6tGEmXj8yF4m+K+Ca+HEmbmdTXXuJ/rQThdv1wA+IvqgnhH7tBVoL1gUxB8fzU1Ov0En2hdW1Zp67H4ofHwbmxY9rrq/xP8k/ADxPjfcrvi3xM+Ao8CPgl8BJj74EHUbXnehL0t8l/hT4AjAcL88hjH458EMz22Zma+N1NT0HJyLRF1zIheHubmY1+XukZtYMPAn8V3c/ZXbuOxhrsV/uPgRcbmYzge8D75viks6bmX0MOOru28xs5VTXU2XXuvsBM5sL/MjMXs3fWItzcCJq7RV6ki+srjVHzGw+QPzn0Xh9zfTVzOqIwvx/ufv34tU13y8Adz8JPEN0K2Jm/CXoMLruC/ol6efhQ8AtZrYXeJzotsufUfv9wt0PxH8eJfoL+EoCmYPjUWuBPvKF1fE78bcTfUF1Lct9wTbxn/+Qt/534nfkrwa68/75+K5h0UvxvwZ2ufsf522q2X6ZWVv8yhwzayJ6T2AXUbB/Im5W2KdcXyf9S9Inyt3vdfd2d19M9Nz5ibv/NjXeLzO7yMym5x4DHwVeoYbn4IRN9U388f4ANwG/ILqn+cWprmectX8HOAScJbpvdyfRPcmngdeAHwOz47ZG9Bs9vwReBjqnuv4SfbqW6P7lduBn8c9NtdwvYAXwUtynV4D74vWXAi8Ae4C/Bxri9Y3x8p54+6VT3YcEfVwJ/CCEfsX1/zz+2ZHLhVqegxP90Uf/RUQCUWu3XEREpAQFuohIIBToIiKBUKCLiARCgS4iEggFuohIIBToIiKB+P+719y8zK6CLAAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xscWoEuBVI3U",
        "colab_type": "code",
        "outputId": "9669fd79-3c77-41c4-cc10-1aad0ddeec63",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        }
      },
      "source": [
        "# Trying to binarise only one column of the training data which is mean area\n",
        "# Finds a threshold a binarizes some of the value to 0 and others to 1\n",
        "#X_binarise_train_mean_area = X_test['mean area'].map(lambda x: 0 if x < 1000 else 1)\n",
        "\n",
        "plt.plot(X_test.T, '*')\n",
        "plt.xticks(rotation = 'vertical')\n",
        "plt.show()"
      ],
      "execution_count": 229,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "RtCYwC97VYfD",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Instead of binarising the columns of the data set manually we can use pandas\n",
        "# to do this for all columns automatically\n",
        "\n",
        "# applying pandas feature function to training data which is \"cut\"\n",
        "# cut splits the particular set of values into bins which we have to specify\n",
        "# and we only want two bins here and I want to say there labels are 0 and 1\n",
        "X_binarised_train = X_train.apply(pd.cut, bins = 3, labels = [0, 1, 2])\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "JaMYQEU7WfvA",
        "colab_type": "code",
        "outputId": "11505adb-5d64-4236-b8e8-d8d03f6960d8",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        }
      },
      "source": [
        "# Let us see whether the binarised training data has been binarised through plot\n",
        "plt.plot(X_binarised_train.T, '*')\n",
        "plt.xticks(rotation = 'vertical')\n",
        "plt.show()"
      ],
      "execution_count": 231,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "RWqmCPmuWsiI",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Instead of binarising the columns of the data set manually we can use pandas\n",
        "# to do this for all columns automatically\n",
        "\n",
        "# applying pandas feature function to test data which is \"cut\"\n",
        "# cut splits the particular set of values into bins which we have to specify\n",
        "# and we only want two bins here and I want to say there labels are 0 and 1\n",
        "X_binarised_test = X_test.apply(pd.cut, bins = 2, labels = [0, 1])\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "gz1cwJOQXcbh",
        "colab_type": "code",
        "outputId": "b38dc80b-1f0d-4ef5-85a2-f58fcf5b30c3",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        }
      },
      "source": [
        "# Let us see whether the binarised training data has been binarised through plot\n",
        "plt.plot(X_binarised_test.T, '*')\n",
        "plt.xticks(rotation = 'vertical')\n",
        "plt.show()"
      ],
      "execution_count": 233,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "YX7bgEI-XkqW",
        "colab_type": "code",
        "outputId": "40d38212-049c-4d87-bf0b-b2a8cfc86daa",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        }
      },
      "source": [
        "print(type(X_binarised_train))\n",
        "print(type(X_binarised_test))"
      ],
      "execution_count": 234,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "<class 'pandas.core.frame.DataFrame'>\n",
            "<class 'pandas.core.frame.DataFrame'>\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bjNimHSTXxtG",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# We want to convert them into array in order for numerical computation \n",
        "# and so we have to convert from pandas data frame to numpy arrays\n",
        "X_binarised_train = X_binarised_train.values\n",
        "X_binarised_test = X_binarised_test.values"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "g9qjohGRYLw_",
        "colab_type": "code",
        "outputId": "43abf319-b395-4b96-9dd5-3f1cb53157be",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        }
      },
      "source": [
        "# Now they are ready to be apllied for applying into the models\n",
        "print(type(X_binarised_train))\n",
        "print(type(X_binarised_test))"
      ],
      "execution_count": 236,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "<class 'numpy.ndarray'>\n",
            "<class 'numpy.ndarray'>\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "_rWyQyUWYrZs",
        "colab_type": "text"
      },
      "source": [
        "# MP neuron model"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "NsRopnHychLK",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Let us see for different values of i\n",
        "# generates random integer\n",
        "\n",
        "from random import randint "
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "eygQTgoyYecl",
        "colab_type": "code",
        "outputId": "735dc821-754b-4950-81ea-4d4229b7d100",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        }
      },
      "source": [
        "# The model of the MP neuron is single parameter (single parametric model)\n",
        "# Given a particular data item (binarized training data set) we add up all the \n",
        "# features which are in the binary format and see if the sum exceeds the value \n",
        "# of the single parameter 'b'\n",
        "# If it exceeds b or equals b then we say the MP neuron model outputs 1 (malignant)\n",
        "# If it does not exceeds then the MP neuron model outputs 0 (begnin)\n",
        "\n",
        "# Let us see with a particular parameter say b as 3\n",
        "\n",
        "# But our goal is to find out the value of b such that the accuracy of finding\n",
        "# the actual value across all the rows by the model is maximized (or) the loss is minimum\n",
        "\n",
        "b = 3 # Let us say the single parameter is initialized as 3\n",
        "\n",
        "#i = 100 # 100 th row of the training data\n",
        "i = randint(0, X_binarised_train.shape[0])\n",
        "print('For row', i)\n",
        "\n",
        "# Predicted value by the model\n",
        "if(np.sum(X_binarised_train[i]) >= b):\n",
        "  print(\"MP neuron inference is malignant\")\n",
        "else:\n",
        "  print(\"MP neuron inference is benign\")\n",
        "\n",
        "# Actual value present in the data set\n",
        "if(Y_train[i] == 1):\n",
        "  print(\"Ground truth is malignant\")\n",
        "else:\n",
        "  print(\"Ground truth is benign\")"
      ],
      "execution_count": 238,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "For row 294\n",
            "MP neuron inference is malignant\n",
            "Ground truth is malignant\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "NinkZAnDeNi_",
        "colab_type": "code",
        "outputId": "815571f0-135f-40d6-c5ec-0f226a4a19c3",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        }
      },
      "source": [
        "# Let us first take a specific value of b and compute the accuracy over all rows\n",
        "b = 3 # initialse b as 3\n",
        "\n",
        "Y_pred_train = [] # empty list or 1D array that will store the predicted values\n",
        "accurate_rows = 0 # keeps the count correctly predicted rows\n",
        "\n",
        "# zip function is used to do two iterations parallely at the same time\n",
        "for x, y in zip(X_binarised_train, Y_train):\n",
        "  # iterating x over X_binarised_train\n",
        "  # iterating y over Y_train\n",
        "  # x is 30 features or one row in X_binarised_train\n",
        "  # y is the associated label or target of that row\n",
        "  # sum of 30 features of a single row compared to b \n",
        "  # which returns a boolean value stored in a variable\n",
        "  y_pred = (np.sum(x) >= b)\n",
        "  # boolean value is appended to the list Y_pred_train\n",
        "  Y_pred_train.append(y_pred)\n",
        "  accurate_rows += y == y_pred\n",
        "# X_binarised_train.shape[0] returns the number of rows value as\n",
        "# X_binarised_train.shape is tuple data type\n",
        "#print(type(y_pred))\n",
        "#print(type(y))\n",
        "print(\"accurate # prediction: \", accurate_rows, \",\", \"accuracy: \", accurate_rows/X_binarised_train.shape[0])\n",
        "# The accuracy is very low because we only get 15% accuracy since the data has \n",
        "# 569 rows out of which only 15% of 569 rows were correctly predicted. This is \n",
        "# because we use the only one bias which is b = 3 that outputs \"malignant\" always.\n",
        "# Actually there are 63 % of data being malignant (refer data.groupby('class').mean() for mean of malignant) out of which only 15% is predicted\n",
        "# correctly when b is 3 \n",
        "''' find this \"data['class'].value_counts()\" and this \"data.describe()\" in this notebook to see the mean value of class malignant'''"
      ],
      "execution_count": 239,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "accurate # prediction:  210 , accuracy:  0.41015625\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "' find this \"data[\\'class\\'].value_counts()\" and this \"data.describe()\" in this notebook to see the mean value of class malignant'"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 239
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "5uTeuvGfjRTf",
        "colab_type": "code",
        "outputId": "a964de07-9c0a-4164-9ac5-64c2c4613c51",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        }
      },
      "source": [
        "# Let us the redo the program for all values of 'b'.\n",
        "# We want to obtain such a value of 'b' such that the accuracy over all the \n",
        "# training dataset is maximized\n",
        "\n",
        "# X_binarised_train.shape[1] returns the # of columns value since it is a \n",
        "# tuple data type and the reason for X_binarised_train.shape[1] + 1 is that \n",
        "# the bias value will also include from 0 to X_binarised_train.shape[1]\n",
        "# or o to # columns of training data + 1\n",
        "# i.e., [0, X_binarised_train.shape[1]]\n",
        "for b in range(X_binarised_train.shape[1] + 1):\n",
        "  Y_pred_train = [] # empty list or 1D array that will store the predicted values\n",
        "  accurate_rows = 0 # keeps the count correctly predicted rows\n",
        "\n",
        "  # zip function is used to do two iterations parallely at the same time\n",
        "  for x, y in zip(X_binarised_train, Y_train):\n",
        "    # iterating x over X_binarised_train\n",
        "    # iterating y over Y_train\n",
        "    # x is 30 features or one row in X_binarised_train\n",
        "    # y is the associated label or target of that row\n",
        "    y_pred = (np.sum(x) >= b) # sum of 30 features of a single row compared to b and stored in a variable\n",
        "    Y_pred_train.append(y_pred) # stored varialbe is appended to the list Y_pred_train\n",
        "    accurate_rows += y == y_pred\n",
        "    # X_binarised_train.shape[0] returns the number of rows value as\n",
        "    # X_binarised_train.shape is tuple data type\n",
        "    # X_binarised_train.shape[0] retuens the # rows\n",
        "  print(\"Bias value: \", b, \", \", \"Accuracy: \", accurate_rows/X_binarised_train.shape[0])\n",
        "  # Accuracy is high when the value of b is 0 and then accuracy is reduced for all other values\n",
        "  # of b for all the training data set\n",
        "  # Actually there are 63 % of data being malignant out of which our MP neuron model manages\n",
        "  # to predict it more or less same but only for bias  value 0\n",
        "  # This means the accuracy reduces again for different values of b and ends up in a situation\n",
        "  # where accuracy drastically reduces and ends up being poor\n",
        "  # Remember our goal is to obtain such a value of 'b' such that the accuracy over all the \n",
        "  # training dataset is maximized and so the model classifies them correctly as possible as actual"
      ],
      "execution_count": 240,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Bias value:  0 ,  Accuracy:  0.626953125\n",
            "Bias value:  1 ,  Accuracy:  0.55859375\n",
            "Bias value:  2 ,  Accuracy:  0.494140625\n",
            "Bias value:  3 ,  Accuracy:  0.41015625\n",
            "Bias value:  4 ,  Accuracy:  0.3046875\n",
            "Bias value:  5 ,  Accuracy:  0.240234375\n",
            "Bias value:  6 ,  Accuracy:  0.154296875\n",
            "Bias value:  7 ,  Accuracy:  0.10546875\n",
            "Bias value:  8 ,  Accuracy:  0.08984375\n",
            "Bias value:  9 ,  Accuracy:  0.091796875\n",
            "Bias value:  10 ,  Accuracy:  0.10546875\n",
            "Bias value:  11 ,  Accuracy:  0.115234375\n",
            "Bias value:  12 ,  Accuracy:  0.12890625\n",
            "Bias value:  13 ,  Accuracy:  0.14453125\n",
            "Bias value:  14 ,  Accuracy:  0.171875\n",
            "Bias value:  15 ,  Accuracy:  0.193359375\n",
            "Bias value:  16 ,  Accuracy:  0.216796875\n",
            "Bias value:  17 ,  Accuracy:  0.236328125\n",
            "Bias value:  18 ,  Accuracy:  0.2578125\n",
            "Bias value:  19 ,  Accuracy:  0.26953125\n",
            "Bias value:  20 ,  Accuracy:  0.2890625\n",
            "Bias value:  21 ,  Accuracy:  0.30859375\n",
            "Bias value:  22 ,  Accuracy:  0.322265625\n",
            "Bias value:  23 ,  Accuracy:  0.326171875\n",
            "Bias value:  24 ,  Accuracy:  0.33203125\n",
            "Bias value:  25 ,  Accuracy:  0.341796875\n",
            "Bias value:  26 ,  Accuracy:  0.3515625\n",
            "Bias value:  27 ,  Accuracy:  0.3515625\n",
            "Bias value:  28 ,  Accuracy:  0.359375\n",
            "Bias value:  29 ,  Accuracy:  0.36328125\n",
            "Bias value:  30 ,  Accuracy:  0.36328125\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "HVm8mePKnjxa",
        "colab_type": "code",
        "outputId": "b4524e89-c640-4dd6-a68b-6cbb4a93309e",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        }
      },
      "source": [
        "# Why does this happen?\n",
        "# Let us look into our data itself to see what needs to be improved to optimize\n",
        "# our model for accurate prediction\n",
        "\n",
        "''' Find this \"data.groupby('class').mean()\" in this notebook to see more about the data'''\n",
        "# What we can vaguely observe in the data is that the features of the data of class 0 (benign) has very high value\n",
        "# when compared to the features with class 1 (malignant)\n",
        "\n",
        "# This means the problem that we face right now to improve the accuracy over all the data points \n",
        "# of the training data seems to be in the way we have binarised the inputs to the model\n",
        "# Remember that class 0 is benign and class 1 is malignant and also we know that the mean of our \n",
        "# binary class label has around 63% of malignant data (All these leads us to make an inference that\n",
        "# the model could improve its accuracy when we try to toggle the labels while binarising the inputs)\n",
        "\n",
        "''' Refer this X_binarised_train = X_train.apply(pd.cut, bins = 2, labels = [0, 1]) and this\n",
        "X_binarised_test = X_test.apply(pd.cut, bins = 2, labels = [0, 1]) to see how the binarisation of inputs or features happen'''\n",
        "# That is whenever we are trying to split the dataset the above code seen within ''' ''' tries to find a threshold value in the features\n",
        "# and assigns o to those values which are below that value and assigns 1 to the rest.\n",
        "# And that tendency to label it will automatically goes to malignant as 1 because we did not observe the data (\n",
        "\n",
        "''' Toggling the labels while binarising the input features'''\n",
        "# But now that we have observed our dataset has very low value for malignant and comparitively high value for benign) we can try to change that\n",
        "# and see whether our model maximizes the accuracy over all the training data set and try to find out such a value of 'b' that maximizes the \n",
        "# accuracy of the model over all different values of bias 'b' in [0, X_binarised_train.shape[1] + 1]\n",
        "\n",
        "# Flipping the labels of feature binarisation for both training and testing data set\n",
        "X_binarised_train = X_train.apply(pd.cut, bins = 2, labels = [1, 0]) # Flipped the labels because we observed different trend from actual dataset\n",
        "X_binarised_test = X_test.apply(pd.cut, bins = 2, labels = [1, 0]) # Flipped the labels because we observed different trend from actual dataset\n",
        "print(type(X_binarised_train))\n",
        "print(type(X_binarised_test))\n",
        "# We want to convert them into array in order for numerical computation \n",
        "# and so we have to convert from pandas data frame to numpy arrays\n",
        "X_binarised_train = X_binarised_train.values\n",
        "X_binarised_test = X_binarised_test.values\n",
        "print(type(X_binarised_train))\n",
        "print(type(X_binarised_test))"
      ],
      "execution_count": 241,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "<class 'pandas.core.frame.DataFrame'>\n",
            "<class 'pandas.core.frame.DataFrame'>\n",
            "<class 'numpy.ndarray'>\n",
            "<class 'numpy.ndarray'>\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "0zHhStR941VD",
        "colab_type": "code",
        "outputId": "7c3c402b-2336-4c59-d720-7808765f11f4",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        }
      },
      "source": [
        "''' Let's re-run the code and see the accuracy of the model over all the training data set for different values of b in an attempt to\n",
        "find out that the value of b which improves the accuracy of classification of the model '''\n",
        "\n",
        "''' Training data (Otimization can be seen as search algorithm and inference (accuracy metrics) problem to find the optimal value) '''\n",
        "# Let us the redo the program for all values of 'b'.\n",
        "# We want to obtain such a value of 'b' such that the accuracy over all the \n",
        "# training dataset is maximized\n",
        "\n",
        "# X_binarised_train.shape[1] returns the # of columns value since it is a \n",
        "# tuple data type and the reason for X_binarised_train.shape[1] + 1 is that \n",
        "# the bias value will also include from 0 to X_binarised_train.shape[1]\n",
        "# or o to # columns of training data + 1\n",
        "# i.e., [0, X_binarised_train.shape[1]]\n",
        "for b in range(X_binarised_train.shape[1] + 1):\n",
        "  Y_pred_train = [] # empty list or 1D array that will store the predicted values\n",
        "  accurate_rows = 0 # keeps the count correctly predicted rows\n",
        "\n",
        "  # zip function is used to do two iterations parallely at the same time\n",
        "  for x, y in zip(X_binarised_train, Y_train):\n",
        "    # iterating x over X_binarised_train\n",
        "    # iterating y over Y_train\n",
        "    # x is 30 features or one row in X_binarised_train\n",
        "    # y is the associated label or target of that row\n",
        "    y_pred_ = (np.sum(x) >= b) # sum of 30 features of a single row compared to b and stored in a variable\n",
        "    Y_pred_train.append(y_pred_) # stored varialbe is appended to the list Y_pred_train\n",
        "    accurate_rows += y == y_pred_\n",
        "    # X_binarised_train.shape[0] returns the number of rows value as\n",
        "    # X_binarised_train.shape is tuple data type\n",
        "    # X_binarised_train.shape[0] retuens the # rows\n",
        "  print(\"Bias value: \", b, \", \", \"Accuracy: \", accurate_rows/X_binarised_train.shape[0])\n",
        "\n",
        "  # So what we see in the result is that the accuracy has consistently increased over the training data set\n",
        "  # and the value of bias or b at 28 gives an accuracy of around 85% which is atleast way better than the previous results\n",
        "  # that was obtained without observing and binarising the trend of the data\n",
        "\n",
        "''' Therefore binarising of the input features plays a key role in increasing the accuracy of the training data '''"
      ],
      "execution_count": 242,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Bias value:  0 ,  Accuracy:  0.626953125\n",
            "Bias value:  1 ,  Accuracy:  0.626953125\n",
            "Bias value:  2 ,  Accuracy:  0.626953125\n",
            "Bias value:  3 ,  Accuracy:  0.626953125\n",
            "Bias value:  4 ,  Accuracy:  0.626953125\n",
            "Bias value:  5 ,  Accuracy:  0.626953125\n",
            "Bias value:  6 ,  Accuracy:  0.626953125\n",
            "Bias value:  7 ,  Accuracy:  0.626953125\n",
            "Bias value:  8 ,  Accuracy:  0.626953125\n",
            "Bias value:  9 ,  Accuracy:  0.626953125\n",
            "Bias value:  10 ,  Accuracy:  0.626953125\n",
            "Bias value:  11 ,  Accuracy:  0.626953125\n",
            "Bias value:  12 ,  Accuracy:  0.626953125\n",
            "Bias value:  13 ,  Accuracy:  0.626953125\n",
            "Bias value:  14 ,  Accuracy:  0.630859375\n",
            "Bias value:  15 ,  Accuracy:  0.6328125\n",
            "Bias value:  16 ,  Accuracy:  0.642578125\n",
            "Bias value:  17 ,  Accuracy:  0.6484375\n",
            "Bias value:  18 ,  Accuracy:  0.65625\n",
            "Bias value:  19 ,  Accuracy:  0.6640625\n",
            "Bias value:  20 ,  Accuracy:  0.671875\n",
            "Bias value:  21 ,  Accuracy:  0.6875\n",
            "Bias value:  22 ,  Accuracy:  0.701171875\n",
            "Bias value:  23 ,  Accuracy:  0.724609375\n",
            "Bias value:  24 ,  Accuracy:  0.755859375\n",
            "Bias value:  25 ,  Accuracy:  0.78515625\n",
            "Bias value:  26 ,  Accuracy:  0.818359375\n",
            "Bias value:  27 ,  Accuracy:  0.845703125\n",
            "Bias value:  28 ,  Accuracy:  0.849609375\n",
            "Bias value:  29 ,  Accuracy:  0.814453125\n",
            "Bias value:  30 ,  Accuracy:  0.759765625\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "' Therefore binarising of the input features plays a key role in increasing the accuracy of the training data '"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 242
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "5JA-IgFj5Ox-",
        "colab_type": "code",
        "outputId": "a40922ce-f45a-4713-ca92-a48da80feb55",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        }
      },
      "source": [
        "''' MP  neuron model - Test data (Inference problem (accuracy metrics)) '''\n",
        "\n",
        "# Let us try to test our MP neuron model accuracy when the bias 'b' is at 28\n",
        "# We can compute accuracy using scikit learn instead of manual computation\n",
        "\n",
        "from sklearn.metrics import accuracy_score\n",
        "b = 28 # the bias that gave highest accuracy for the model in the training data\n",
        "Y_pred_test = [] # empty list or 1D array\n",
        "for x in X_binarised_test: # iterate over all each test data\n",
        "  y_pred = (np.sum(x) >= b) # y_pred stores the scalar value summation of 30 features computed and compared with b value\n",
        "  Y_pred_test.append(y_pred) # stores the y_pred value computed as a list for each test data\n",
        "  ''' The function of accuracy_score is to return a float value corresponding to the\n",
        "  similarities between true labels and predicted labels'''\n",
        "accuracy = accuracy_score(Y_pred_test, Y_test) # computes the accuracy\n",
        "print(\"Bias: \", b, \",\", \"Accuracy: \", accuracy)\n",
        "\n",
        "# What we can observe here is that the accuracy of the MP neuron model on the test data is low when compared to\n",
        "# the accuracy acquired in the training data\n",
        "# This is not in our control (we can only train our model through training data and try to search and infer the best value (optimal value) of\n",
        "# b or bias for training data alone. But the behavior that this bias or b exhibits for the test data is not in our hand for improvisation\n",
        "# i.e., we have control only over training data so that we can try to improvise our models' accuracy whereas in the testing part we are\n",
        "# only inferring how the trained model behaves or how much accuracy does it exhibits for unseen data)"
      ],
      "execution_count": 243,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Bias:  28 , Accuracy:  0.7894736842105263\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "1BeR-3NMDNod",
        "colab_type": "text"
      },
      "source": [
        "# Practise session"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "HwQ6do89Fj0P",
        "colab_type": "code",
        "outputId": "37653ae0-b906-48c9-c1f9-4d71f4a5d502",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 785
        }
      },
      "source": [
        "#for x, y in zip(X_binarised_train, Y_train)\n",
        "#bias = 5\n",
        "\n",
        "# inference and search problem\n",
        "for bias in range(X_binarised_train.shape[1] + 1):\n",
        "  pred_acc_rows = 0\n",
        "  mod_pred_train = []\n",
        "  for row, label in zip(X_binarised_train, Y_train):\n",
        "    mod_pred_val = (np.sum(row) >= bias)\n",
        "    mod_pred_train.append(mod_pred_val)\n",
        "    pred_acc_rows += label == mod_pred_val\n",
        "  #print(\"Accurate rows predicted: \", pred_acc_rows)  \n",
        "  print(\"Bias:\", bias, \",\", \"Accurate rows predicted:\", pred_acc_rows, \",\", \n",
        "        \"Accuracy:\", pred_acc_rows/X_binarised_train.shape[0])\n",
        "print()\n",
        "print(type(X_binarised_train.shape))\n",
        "print(type(Y_train.shape))\n",
        "print(type(label))\n",
        "print(type(mod_pred_val))\n",
        "print(mod_pred_train)\n",
        "\n",
        "print(mod_pred_train.index(False))\n",
        "print(mod_pred_train.index(True))\n",
        "print(type(mod_pred_train.index))\n",
        "\n",
        "index?\n",
        "#list?\n",
        "\n",
        "\n",
        "# False indices\n",
        "\n",
        "print (\"The original list is : \", mod_pred_train)\n",
        "\n",
        "# lambda + filter() + range()  \n",
        "\n",
        "false_indices = list(filter(lambda i: not mod_pred_train[i], range(len(mod_pred_train))))\n",
        "\n",
        "print (\"List of indices with False values: \", false_indices)\n",
        "\n",
        "\n",
        "\n",
        "# True indices\n",
        "\n",
        "print (\"The original list is : \", mod_pred_train)\n",
        "\n",
        "# lambda + filter() + range() \n",
        "\n",
        "true_indices = list(filter(lambda i: mod_pred_train[i], range(len(mod_pred_train))))\n",
        "\n",
        "print (\"List of indices with True values: \", true_indices)\n",
        "\n",
        "\n",
        "\n",
        "\n",
        "\n",
        "#print(mod_pred_train.index('False'))\n",
        "#from itertools import compress  \n",
        "\n",
        "#print (\"Given list is : \" +  str(mod_pred_train))  \n",
        "#res = list(compress(range(len(mod_pred_train)), mod_pred_train ))  \n",
        "#print (\"Indices having True values are : \" +  str(res))\n",
        "\n",
        "\n",
        "#print(type(pred_acc))\n",
        "#print(pred_acc)\n",
        "#print(type(Y_train))  \n",
        "#print(type(mod_pred))\n",
        "#print(pred_acc)\n",
        "#print(type(accurate_rows), type(pred_acc))\n",
        "#accurate_rows += y == y_pred_"
      ],
      "execution_count": 244,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Bias: 0 , Accurate rows predicted: 321 , Accuracy: 0.626953125\n",
            "Bias: 1 , Accurate rows predicted: 321 , Accuracy: 0.626953125\n",
            "Bias: 2 , Accurate rows predicted: 321 , Accuracy: 0.626953125\n",
            "Bias: 3 , Accurate rows predicted: 321 , Accuracy: 0.626953125\n",
            "Bias: 4 , Accurate rows predicted: 321 , Accuracy: 0.626953125\n",
            "Bias: 5 , Accurate rows predicted: 321 , Accuracy: 0.626953125\n",
            "Bias: 6 , Accurate rows predicted: 321 , Accuracy: 0.626953125\n",
            "Bias: 7 , Accurate rows predicted: 321 , Accuracy: 0.626953125\n",
            "Bias: 8 , Accurate rows predicted: 321 , Accuracy: 0.626953125\n",
            "Bias: 9 , Accurate rows predicted: 321 , Accuracy: 0.626953125\n",
            "Bias: 10 , Accurate rows predicted: 321 , Accuracy: 0.626953125\n",
            "Bias: 11 , Accurate rows predicted: 321 , Accuracy: 0.626953125\n",
            "Bias: 12 , Accurate rows predicted: 321 , Accuracy: 0.626953125\n",
            "Bias: 13 , Accurate rows predicted: 321 , Accuracy: 0.626953125\n",
            "Bias: 14 , Accurate rows predicted: 323 , Accuracy: 0.630859375\n",
            "Bias: 15 , Accurate rows predicted: 324 , Accuracy: 0.6328125\n",
            "Bias: 16 , Accurate rows predicted: 329 , Accuracy: 0.642578125\n",
            "Bias: 17 , Accurate rows predicted: 332 , Accuracy: 0.6484375\n",
            "Bias: 18 , Accurate rows predicted: 336 , Accuracy: 0.65625\n",
            "Bias: 19 , Accurate rows predicted: 340 , Accuracy: 0.6640625\n",
            "Bias: 20 , Accurate rows predicted: 344 , Accuracy: 0.671875\n",
            "Bias: 21 , Accurate rows predicted: 352 , Accuracy: 0.6875\n",
            "Bias: 22 , Accurate rows predicted: 359 , Accuracy: 0.701171875\n",
            "Bias: 23 , Accurate rows predicted: 371 , Accuracy: 0.724609375\n",
            "Bias: 24 , Accurate rows predicted: 387 , Accuracy: 0.755859375\n",
            "Bias: 25 , Accurate rows predicted: 402 , Accuracy: 0.78515625\n",
            "Bias: 26 , Accurate rows predicted: 419 , Accuracy: 0.818359375\n",
            "Bias: 27 , Accurate rows predicted: 433 , Accuracy: 0.845703125\n",
            "Bias: 28 , Accurate rows predicted: 435 , Accuracy: 0.849609375\n",
            "Bias: 29 , Accurate rows predicted: 417 , Accuracy: 0.814453125\n",
            "Bias: 30 , Accurate rows predicted: 389 , Accuracy: 0.759765625\n",
            "\n",
            "<class 'tuple'>\n",
            "<class 'tuple'>\n",
            "<class 'int'>\n",
            "<class 'bool'>\n",
            "[False, True, False, True, True, False, True, False, False, False, False, False, True, True, False, True, False, False, True, True, False, False, True, True, False, False, False, True, True, False, False, False, False, False, False, True, True, False, False, True, False, False, False, True, False, False, True, False, False, True, True, True, True, False, False, True, False, False, True, False, False, False, True, True, False, False, False, True, True, False, True, False, False, True, False, True, False, False, True, False, False, False, False, True, False, False, False, True, False, True, True, False, False, True, False, False, False, False, True, False, True, False, False, False, True, True, False, False, False, True, False, False, False, False, True, False, True, True, True, False, False, True, False, False, True, False, False, False, False, True, True, True, True, True, True, False, False, True, True, True, False, False, False, False, True, False, False, True, True, False, False, False, False, True, True, True, False, True, True, True, False, True, False, False, True, True, False, False, False, False, True, False, True, False, False, False, False, True, True, False, True, False, False, True, True, True, False, False, True, True, False, False, False, True, False, False, True, False, True, True, False, True, False, True, True, False, False, True, False, True, True, True, True, True, True, False, False, True, True, False, False, False, True, False, True, True, True, True, False, False, False, False, True, True, False, False, False, False, True, False, True, False, False, False, True, True, False, False, False, False, True, False, False, True, True, False, True, False, False, False, True, False, True, True, False, True, False, True, False, False, False, False, True, True, True, False, False, False, False, False, True, True, False, True, True, False, False, True, True, False, False, True, False, True, True, True, True, True, False, False, True, False, False, True, True, False, True, False, True, False, True, False, True, False, True, False, False, False, True, False, True, False, True, False, False, True, True, False, False, False, False, False, False, False, False, False, False, True, False, False, False, False, False, False, False, False, False, True, True, False, False, False, True, False, False, True, False, True, True, False, True, True, True, True, False, True, False, False, False, True, True, False, False, True, False, True, False, False, False, False, False, True, False, False, False, True, True, False, False, True, True, True, True, False, False, True, False, True, True, False, False, True, False, False, True, True, True, False, False, True, False, True, False, False, False, False, False, False, False, False, True, True, False, True, True, False, True, True, True, True, False, True, True, False, False, True, False, False, False, False, True, True, True, False, False, False, True, True, True, False, False, False, True, True, True, True, False, False, True, False, False, True, False, True, False, False, False, False, False, False, True, False, False, False, False, False, False, False, True, True, False, False, True, True, True, True, True, True, False, False, False, True, True, False, True, False, True, True, False, True, True, False, False, False, True, False, False, False, True, True, False, False]\n",
            "0\n",
            "1\n",
            "<class 'builtin_function_or_method'>\n",
            "The original list is :  [False, True, False, True, True, False, True, False, False, False, False, False, True, True, False, True, False, False, True, True, False, False, True, True, False, False, False, True, True, False, False, False, False, False, False, True, True, False, False, True, False, False, False, True, False, False, True, False, False, True, True, True, True, False, False, True, False, False, True, False, False, False, True, True, False, False, False, True, True, False, True, False, False, True, False, True, False, False, True, False, False, False, False, True, False, False, False, True, False, True, True, False, False, True, False, False, False, False, True, False, True, False, False, False, True, True, False, False, False, True, False, False, False, False, True, False, True, True, True, False, False, True, False, False, True, False, False, False, False, True, True, True, True, True, True, False, False, True, True, True, False, False, False, False, True, False, False, True, True, False, False, False, False, True, True, True, False, True, True, True, False, True, False, False, True, True, False, False, False, False, True, False, True, False, False, False, False, True, True, False, True, False, False, True, True, True, False, False, True, True, False, False, False, True, False, False, True, False, True, True, False, True, False, True, True, False, False, True, False, True, True, True, True, True, True, False, False, True, True, False, False, False, True, False, True, True, True, True, False, False, False, False, True, True, False, False, False, False, True, False, True, False, False, False, True, True, False, False, False, False, True, False, False, True, True, False, True, False, False, False, True, False, True, True, False, True, False, True, False, False, False, False, True, True, True, False, False, False, False, False, True, True, False, True, True, False, False, True, True, False, False, True, False, True, True, True, True, True, False, False, True, False, False, True, True, False, True, False, True, False, True, False, True, False, True, False, False, False, True, False, True, False, True, False, False, True, True, False, False, False, False, False, False, False, False, False, False, True, False, False, False, False, False, False, False, False, False, True, True, False, False, False, True, False, False, True, False, True, True, False, True, True, True, True, False, True, False, False, False, True, True, False, False, True, False, True, False, False, False, False, False, True, False, False, False, True, True, False, False, True, True, True, True, False, False, True, False, True, True, False, False, True, False, False, True, True, True, False, False, True, False, True, False, False, False, False, False, False, False, False, True, True, False, True, True, False, True, True, True, True, False, True, True, False, False, True, False, False, False, False, True, True, True, False, False, False, True, True, True, False, False, False, True, True, True, True, False, False, True, False, False, True, False, True, False, False, False, False, False, False, True, False, False, False, False, False, False, False, True, True, False, False, True, True, True, True, True, True, False, False, False, True, True, False, True, False, True, True, False, True, True, False, False, False, True, False, False, False, True, True, False, False]\n",
            "List of indices with False values:  [0, 2, 5, 7, 8, 9, 10, 11, 14, 16, 17, 20, 21, 24, 25, 26, 29, 30, 31, 32, 33, 34, 37, 38, 40, 41, 42, 44, 45, 47, 48, 53, 54, 56, 57, 59, 60, 61, 64, 65, 66, 69, 71, 72, 74, 76, 77, 79, 80, 81, 82, 84, 85, 86, 88, 91, 92, 94, 95, 96, 97, 99, 101, 102, 103, 106, 107, 108, 110, 111, 112, 113, 115, 119, 120, 122, 123, 125, 126, 127, 128, 135, 136, 140, 141, 142, 143, 145, 146, 149, 150, 151, 152, 156, 160, 162, 163, 166, 167, 168, 169, 171, 173, 174, 175, 176, 179, 181, 182, 186, 187, 190, 191, 192, 194, 195, 197, 200, 202, 205, 206, 208, 215, 216, 219, 220, 221, 223, 228, 229, 230, 231, 234, 235, 236, 237, 239, 241, 242, 243, 246, 247, 248, 249, 251, 252, 255, 257, 258, 259, 261, 264, 266, 268, 269, 270, 271, 275, 276, 277, 278, 279, 282, 285, 286, 289, 290, 292, 298, 299, 301, 302, 305, 307, 309, 311, 313, 315, 316, 317, 319, 321, 323, 324, 327, 328, 329, 330, 331, 332, 333, 334, 335, 336, 338, 339, 340, 341, 342, 343, 344, 345, 346, 349, 350, 351, 353, 354, 356, 359, 364, 366, 367, 368, 371, 372, 374, 376, 377, 378, 379, 380, 382, 383, 384, 387, 388, 393, 394, 396, 399, 400, 402, 403, 407, 408, 410, 412, 413, 414, 415, 416, 417, 418, 419, 422, 425, 430, 433, 434, 436, 437, 438, 439, 443, 444, 445, 449, 450, 451, 456, 457, 459, 460, 462, 464, 465, 466, 467, 468, 469, 471, 472, 473, 474, 475, 476, 477, 480, 481, 488, 489, 490, 493, 495, 498, 501, 502, 503, 505, 506, 507, 510, 511]\n",
            "The original list is :  [False, True, False, True, True, False, True, False, False, False, False, False, True, True, False, True, False, False, True, True, False, False, True, True, False, False, False, True, True, False, False, False, False, False, False, True, True, False, False, True, False, False, False, True, False, False, True, False, False, True, True, True, True, False, False, True, False, False, True, False, False, False, True, True, False, False, False, True, True, False, True, False, False, True, False, True, False, False, True, False, False, False, False, True, False, False, False, True, False, True, True, False, False, True, False, False, False, False, True, False, True, False, False, False, True, True, False, False, False, True, False, False, False, False, True, False, True, True, True, False, False, True, False, False, True, False, False, False, False, True, True, True, True, True, True, False, False, True, True, True, False, False, False, False, True, False, False, True, True, False, False, False, False, True, True, True, False, True, True, True, False, True, False, False, True, True, False, False, False, False, True, False, True, False, False, False, False, True, True, False, True, False, False, True, True, True, False, False, True, True, False, False, False, True, False, False, True, False, True, True, False, True, False, True, True, False, False, True, False, True, True, True, True, True, True, False, False, True, True, False, False, False, True, False, True, True, True, True, False, False, False, False, True, True, False, False, False, False, True, False, True, False, False, False, True, True, False, False, False, False, True, False, False, True, True, False, True, False, False, False, True, False, True, True, False, True, False, True, False, False, False, False, True, True, True, False, False, False, False, False, True, True, False, True, True, False, False, True, True, False, False, True, False, True, True, True, True, True, False, False, True, False, False, True, True, False, True, False, True, False, True, False, True, False, True, False, False, False, True, False, True, False, True, False, False, True, True, False, False, False, False, False, False, False, False, False, False, True, False, False, False, False, False, False, False, False, False, True, True, False, False, False, True, False, False, True, False, True, True, False, True, True, True, True, False, True, False, False, False, True, True, False, False, True, False, True, False, False, False, False, False, True, False, False, False, True, True, False, False, True, True, True, True, False, False, True, False, True, True, False, False, True, False, False, True, True, True, False, False, True, False, True, False, False, False, False, False, False, False, False, True, True, False, True, True, False, True, True, True, True, False, True, True, False, False, True, False, False, False, False, True, True, True, False, False, False, True, True, True, False, False, False, True, True, True, True, False, False, True, False, False, True, False, True, False, False, False, False, False, False, True, False, False, False, False, False, False, False, True, True, False, False, True, True, True, True, True, True, False, False, False, True, True, False, True, False, True, True, False, True, True, False, False, False, True, False, False, False, True, True, False, False]\n",
            "List of indices with True values:  [1, 3, 4, 6, 12, 13, 15, 18, 19, 22, 23, 27, 28, 35, 36, 39, 43, 46, 49, 50, 51, 52, 55, 58, 62, 63, 67, 68, 70, 73, 75, 78, 83, 87, 89, 90, 93, 98, 100, 104, 105, 109, 114, 116, 117, 118, 121, 124, 129, 130, 131, 132, 133, 134, 137, 138, 139, 144, 147, 148, 153, 154, 155, 157, 158, 159, 161, 164, 165, 170, 172, 177, 178, 180, 183, 184, 185, 188, 189, 193, 196, 198, 199, 201, 203, 204, 207, 209, 210, 211, 212, 213, 214, 217, 218, 222, 224, 225, 226, 227, 232, 233, 238, 240, 244, 245, 250, 253, 254, 256, 260, 262, 263, 265, 267, 272, 273, 274, 280, 281, 283, 284, 287, 288, 291, 293, 294, 295, 296, 297, 300, 303, 304, 306, 308, 310, 312, 314, 318, 320, 322, 325, 326, 337, 347, 348, 352, 355, 357, 358, 360, 361, 362, 363, 365, 369, 370, 373, 375, 381, 385, 386, 389, 390, 391, 392, 395, 397, 398, 401, 404, 405, 406, 409, 411, 420, 421, 423, 424, 426, 427, 428, 429, 431, 432, 435, 440, 441, 442, 446, 447, 448, 452, 453, 454, 455, 458, 461, 463, 470, 478, 479, 482, 483, 484, 485, 486, 487, 491, 492, 494, 496, 497, 499, 500, 504, 508, 509]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qLErCfqpbhrY",
        "colab_type": "code",
        "outputId": "2ce3dc14-11a8-48b0-992d-e6e8773c3111",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "print(type(X_binarised_train.shape))\n",
        "print(type(Y_train.shape))\n",
        "#print(type(Y_train))\n",
        "Z_train = Y_train\n",
        "#print(type(Z_train))\n",
        "Z_train = Z_train.values\n",
        "#print(type(Z_train))\n",
        "#print(Z_train)\n",
        "#print(Z_train[0], Z_train[1], Z_train[2], Z_train[3], Z_train[4], Z_train[5])\n",
        "\n",
        "count_1 = 0\n",
        "for index, i in enumerate(Z_train):\n",
        "  if (i == 1):\n",
        "    #print(index)\n",
        "    count_1 += 1\n",
        "print()\n",
        "print(\"# 1's: \", count_1, \"# 0's: \", Z_train.shape[0] - count_1)\n",
        "count_0 = 0\n",
        "for index, i in enumerate(Z_train):\n",
        "  if (i == 0):\n",
        "    #print(index)\n",
        "    count_0 += 1\n",
        "print()\n",
        "print(\"# 0's: \", count_0, \"# 1's: \", Z_train.shape[0] - count_0)\n",
        "print()\n",
        "print(count_0, count_1, count_0 + count_1, Z_train.shape[0])\n",
        "\n",
        "\n",
        "''' int <- int + bool '''\n",
        "print()\n",
        "f = 5\n",
        "y_t = True # 1\n",
        "y_f = False # 0\n",
        "\n",
        "z_t = f + y_t\n",
        "print(y_t, type(y_t))\n",
        "print(z_t, type(z_t))\n",
        "\n",
        "z_f = f + y_f\n",
        "print(y_f, type(y_f))\n",
        "print(z_f, type(z_f))\n",
        "\n",
        "\n",
        "print()\n",
        "\n",
        "#for i in X_binarised_train[0]:\n",
        "#  print(i)\n",
        "\n",
        "#print(X_binarised_train[0])\n",
        "#print(Y_train.head())\n",
        "#print()\n",
        "#print(Y_train[430], Y_train[48], Y_train[467], Y_train[547])\n",
        "#print(X_binarised_train[430])\n",
        "#print()\n",
        "#print(Y_train[0], Y_train[1], Y_train[2], Y_train[3], Y_train[4])\n",
        "#print(Y_train[0:10])\n",
        "#print(Y_train[1])\n",
        "#print(type(Y_train))\n",
        "#print(type(X_binarised_train))\n",
        "\n",
        "for row in X_binarised_train:\n",
        "  print(row)\n",
        "\n",
        "## iterating over series\n",
        "for items in Y_train.iteritems():\n",
        "  print(items)"
      ],
      "execution_count": 245,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "<class 'tuple'>\n",
            "<class 'tuple'>\n",
            "\n",
            "# 1's:  321 # 0's:  191\n",
            "\n",
            "# 0's:  191 # 1's:  321\n",
            "\n",
            "191 321 512 512\n",
            "\n",
            "True <class 'bool'>\n",
            "6 <class 'int'>\n",
            "False <class 'bool'>\n",
            "5 <class 'int'>\n",
            "\n",
            "[1 1 1 1 1 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 0 1 1 0 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 0 1 0]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[0 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[0 0 0 1 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 0 0 0 1 0 0 0 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 0 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 0 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[0 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[0 0 0 0 0 0 0 0 1 1 0 1 0 0 1 1 1 1 1 1 0 0 0 0 1 1 0 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[0 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[0 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 1 1 1 1 1 1 1 1]\n",
            "[0 0 0 1 0 1 1 0 1 1 1 0 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 0 1 0 0 1 1 1 1 1 0 0 0 1 1 1 1 1 1 1 1 0 0 0 1]\n",
            "[1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 1 1]\n",
            "[1 1 1 1 0 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1]\n",
            "[0 1 0 1 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 0 1 0 1 0 0 0 0 0 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 1 1]\n",
            "[0 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 0 1 1 1 1 0 0 1 1 1 0 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1]\n",
            "[0 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 0 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[0 1 0 1 0 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 1 1 1 1 0 1 1]\n",
            "[0 1 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 0 1 0 1 1 0 0 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1]\n",
            "[0 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 0 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 0 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1]\n",
            "[1 1 1 1 0 1 1 1 0 0 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 0 0 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 0 1 1 0 0 1 0 1 1 1 0 1 1 0 1 1 0 1 1 1 0 0 0 0 0]\n",
            "[1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 1 1]\n",
            "[0 1 0 1 1 0 0 0 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 0 1 1 0 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 0 0 0 0 0]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[0 1 0 1 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 1 1]\n",
            "[0 1 0 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 0 1 0 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 0 1 0 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 0 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 0 0 0 1]\n",
            "[0 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 0 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1]\n",
            "[0 1 0 0 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 0 1 1 1 0 1 1]\n",
            "[0 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 0 1 1 1 0 0 1 0 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 0 0 0 1 0]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 0 1 0 1 1 1 0 0 1 1]\n",
            "[1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 0 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 0 1]\n",
            "[1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1]\n",
            "[0 1 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1]\n",
            "[0 1 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 0 1 1 1 0 1 1]\n",
            "[0 1 0 0 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[0 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 0 1 1 1 1 1 1 0 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[0 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 1 0 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[0 1 0 1 0 1 0 0 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 0 1 1 0 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 0 0 0 1 1 0 1 1 1 1 1 0 1 1 1 1 1 0 1 1 0 0 0 0 0 0]\n",
            "[1 1 1 1 0 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 1 1 0 1 1]\n",
            "[1 1 1 1 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 0 1 0 1 0 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 0 0 0 0 0 0 1 1 1 1 1 1 1 1 0 1 1 1 1 1 0 1 0 0 0 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 0 1 1]\n",
            "[0 1 0 1 1 1 0 0 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 1 0 0 0 0 1 1]\n",
            "[0 1 0 1 0 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 0 1 0 0 1 1]\n",
            "[1 1 1 1 0 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[0 1 0 0 0 0 0 0 0 1 1 1 1 1 1 0 1 1 1 1 0 1 0 0 0 0 0 0 1 1]\n",
            "[1 0 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 0 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 0 1 1 0 1 1 1 1 1]\n",
            "[0 1 0 1 1 1 1 0 0 1 1 1 1 1 1 1 1 1 1 1 0 1 0 1 1 1 1 0 1 1]\n",
            "[1 0 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 0 0 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 0 1 0 0 1 0 1 1 1 0 0 0 1 0 1 1 1 1 1 1 0 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 1 0 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1]\n",
            "[1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1]\n",
            "[0 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 1 1 0 1 0 1 1]\n",
            "[0 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 0 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 1 1]\n",
            "[1 1 1 1 0 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[0 1 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 0 1 0 0 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 0 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[0 1 0 1 0 1 0 0 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 1 1 1 1 0 1 1]\n",
            "[0 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[0 0 0 1 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 0 0 0 1 1 0 0 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 0]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 0 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 0 1 1]\n",
            "[1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[0 1 0 0 0 1 0 0 1 1 0 1 0 0 1 1 1 1 0 1 0 1 0 0 1 1 1 0 1 1]\n",
            "[1 0 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 1 0 0 0 0 1 0]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 0 1 1 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[0 1 0 1 0 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[0 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 0 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 1 1]\n",
            "[1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 1 0 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 0 0 0 0 0 0 1 0 1 1 1 1 1 0 1 1 1 0 1 1 1 1 1 0 1 1]\n",
            "[0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 0 1 1 1 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1]\n",
            "[1 1 1 1 0 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 1 1]\n",
            "[1 1 1 1 0 1 1 0 1 1 1 0 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 0 1 1]\n",
            "[0 1 0 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 0 1 1 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[0 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 0 0 1 1 0 0 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 0 0 0 0 0]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 0 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1]\n",
            "[0 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 1 1 1 1 0 1 1]\n",
            "[0 1 0 0 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 1 1 1 1 0 1 1]\n",
            "[0 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 1 0 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 0 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[0 0 0 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 0 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[0 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[0 1 0 1 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 0 0 0 1 0 1 1 0 0 1]\n",
            "[1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 0 1 0 0 1 1 1 1 1 0 1 0 1 1 1 1 1 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[0 1 0 0 1 1 0 0 0 1 1 1 1 1 1 1 1 1 1 1 0 1 0 1 1 1 0 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 1 1]\n",
            "[1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 1 1]\n",
            "[0 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 0 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[0 1 0 1 1 0 1 0 0 1 1 1 1 1 1 0 1 1 0 1 0 0 0 1 1 0 0 0 0 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1]\n",
            "[0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[0 0 0 0 1 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 0 0 1 1 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1]\n",
            "[1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 0 0 1 1 1 1 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1]\n",
            "[1 0 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 0 0 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 0 0 0 1 0]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[0 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 0 0 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 0 0 0 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 1 1 0 1 1]\n",
            "[1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 1 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 0 1 1 0 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1]\n",
            "[0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1]\n",
            "[0 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 0 1]\n",
            "[0 1 0 1 1 1 1 0 1 1 1 0 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 0 1 1]\n",
            "[0 1 0 0 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 0 1 1 1 0 1 1]\n",
            "[1 0 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 1 1 1 1 1]\n",
            "[1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 1 1 1 1 0 1 1]\n",
            "[0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 0 1 1 1 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 1 1]\n",
            "[1 1 1 1 0 0 1 1 1 0 1 1 1 1 0 1 1 1 1 1 1 1 1 1 0 1 1 1 1 0]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[0 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 0 1]\n",
            "[1 1 1 1 0 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[0 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 1 1]\n",
            "[1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 0 1 0 1 1 1 0 1 1 1 0 1 1 1 1 1 1 1 1 1 1]\n",
            "[0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 0 0 0 1 0 1 0 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 0 1 1 0 1 1 0 1 1 0 0 1]\n",
            "[1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1]\n",
            "[1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 0 0 1 0 0 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[0 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[0 1 0 0 1 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 0 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[0 1 0 0 0 1 0 0 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 0 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 0 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[0 1 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 1 0 1 1 0 1 1]\n",
            "[0 0 0 0 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1]\n",
            "[0 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 0 1 1 1 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 0 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 0 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[0 1 0 0 1 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 0 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 0 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 0 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[0 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 0 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 0 1 1 0 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 0 1 1]\n",
            "[1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1]\n",
            "[0 1 0 1 0 0 0 0 0 0 1 1 1 1 1 1 1 1 0 1 0 0 0 1 0 0 0 0 0 1]\n",
            "[0 1 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 0 1 1 1 0 0 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 1 1 0 0 1]\n",
            "[1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 0 1 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[0 0 0 1 1 1 1 0 1 1 1 0 1 1 1 1 1 1 1 1 0 0 0 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 0 0 0 0 0 0 1 1 1 1 1 0 1 1 0 1 1 1 1 1 0 0 0 0 0 0]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[0 0 0 1 1 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 0 0 0 1 1 1 1 0 1 1]\n",
            "[0 0 0 1 1 1 0 0 0 1 1 1 1 1 1 1 1 1 1 1 0 0 1 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[0 1 0 1 0 1 0 0 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1]\n",
            "[1 0 1 1 1 1 1 1 1 1 1 1 1 1 0 0 1 0 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[0 1 0 1 1 1 0 0 0 0 1 1 1 1 1 1 1 1 1 1 0 1 0 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[0 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1]\n",
            "[0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 0 1 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 1 1]\n",
            "[0 1 0 1 1 0 1 0 0 0 1 0 1 1 1 0 1 0 0 1 1 1 0 1 1 1 1 0 1 1]\n",
            "[1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1]\n",
            "[0 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 0 0 1 1 0 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 0 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1]\n",
            "[1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 0 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 0 1 0 0 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 0 0 1 1 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 0 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 0 1]\n",
            "[1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 1 1]\n",
            "[0 1 0 1 0 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[0 1 0 1 1 0 1 0 0 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 0 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 0 1 1 1 0 0 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 0 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 1 1 0 1 0 1 1]\n",
            "[1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 0 0 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
            "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1]\n",
            "[0 1 0 1 1 1 1 0 1 1 1 1 1 1 1 1 1 0 1 1 0 1 1 1 1 1 1 0 1 1]\n",
            "(430, 0)\n",
            "(48, 1)\n",
            "(105, 0)\n",
            "(467, 1)\n",
            "(547, 1)\n",
            "(365, 0)\n",
            "(295, 1)\n",
            "(567, 0)\n",
            "(297, 0)\n",
            "(172, 0)\n",
            "(506, 1)\n",
            "(488, 1)\n",
            "(115, 1)\n",
            "(466, 1)\n",
            "(218, 0)\n",
            "(74, 1)\n",
            "(461, 0)\n",
            "(484, 1)\n",
            "(79, 1)\n",
            "(226, 1)\n",
            "(37, 1)\n",
            "(44, 0)\n",
            "(324, 1)\n",
            "(13, 0)\n",
            "(277, 0)\n",
            "(134, 0)\n",
            "(83, 0)\n",
            "(153, 1)\n",
            "(327, 1)\n",
            "(68, 1)\n",
            "(392, 0)\n",
            "(61, 1)\n",
            "(0, 0)\n",
            "(435, 0)\n",
            "(70, 0)\n",
            "(438, 1)\n",
            "(284, 1)\n",
            "(305, 1)\n",
            "(71, 1)\n",
            "(524, 1)\n",
            "(106, 1)\n",
            "(282, 0)\n",
            "(349, 1)\n",
            "(483, 1)\n",
            "(2, 0)\n",
            "(393, 0)\n",
            "(307, 1)\n",
            "(414, 0)\n",
            "(27, 0)\n",
            "(58, 1)\n",
            "(388, 1)\n",
            "(347, 1)\n",
            "(500, 1)\n",
            "(64, 0)\n",
            "(440, 1)\n",
            "(395, 1)\n",
            "(101, 1)\n",
            "(103, 1)\n",
            "(514, 0)\n",
            "(63, 1)\n",
            "(109, 1)\n",
            "(146, 0)\n",
            "(50, 1)\n",
            "(92, 1)\n",
            "(190, 0)\n",
            "(131, 0)\n",
            "(468, 0)\n",
            "(384, 1)\n",
            "(495, 1)\n",
            "(15, 0)\n",
            "(419, 1)\n",
            "(400, 0)\n",
            "(85, 0)\n",
            "(124, 1)\n",
            "(288, 1)\n",
            "(224, 1)\n",
            "(566, 0)\n",
            "(228, 1)\n",
            "(303, 1)\n",
            "(473, 1)\n",
            "(465, 1)\n",
            "(370, 0)\n",
            "(516, 0)\n",
            "(147, 1)\n",
            "(557, 1)\n",
            "(503, 0)\n",
            "(444, 0)\n",
            "(551, 1)\n",
            "(502, 1)\n",
            "(158, 1)\n",
            "(88, 1)\n",
            "(489, 0)\n",
            "(427, 1)\n",
            "(36, 0)\n",
            "(562, 0)\n",
            "(345, 1)\n",
            "(563, 0)\n",
            "(445, 1)\n",
            "(222, 1)\n",
            "(512, 0)\n",
            "(287, 1)\n",
            "(35, 0)\n",
            "(248, 1)\n",
            "(180, 0)\n",
            "(292, 1)\n",
            "(429, 1)\n",
            "(549, 1)\n",
            "(368, 0)\n",
            "(449, 0)\n",
            "(278, 1)\n",
            "(186, 0)\n",
            "(528, 1)\n",
            "(56, 0)\n",
            "(132, 0)\n",
            "(276, 1)\n",
            "(162, 0)\n",
            "(464, 1)\n",
            "(364, 1)\n",
            "(434, 1)\n",
            "(57, 0)\n",
            "(39, 0)\n",
            "(309, 1)\n",
            "(9, 0)\n",
            "(114, 1)\n",
            "(299, 1)\n",
            "(193, 0)\n",
            "(25, 0)\n",
            "(16, 0)\n",
            "(351, 0)\n",
            "(361, 1)\n",
            "(273, 1)\n",
            "(21, 1)\n",
            "(508, 1)\n",
            "(313, 1)\n",
            "(246, 1)\n",
            "(253, 0)\n",
            "(126, 0)\n",
            "(423, 1)\n",
            "(220, 1)\n",
            "(333, 1)\n",
            "(182, 0)\n",
            "(252, 0)\n",
            "(45, 0)\n",
            "(501, 0)\n",
            "(332, 1)\n",
            "(108, 0)\n",
            "(38, 0)\n",
            "(173, 1)\n",
            "(497, 1)\n",
            "(80, 1)\n",
            "(535, 0)\n",
            "(555, 1)\n",
            "(168, 0)\n",
            "(84, 1)\n",
            "(184, 0)\n",
            "(360, 1)\n",
            "(152, 1)\n",
            "(169, 1)\n",
            "(111, 1)\n",
            "(312, 1)\n",
            "(460, 0)\n",
            "(425, 1)\n",
            "(526, 1)\n",
            "(452, 1)\n",
            "(544, 1)\n",
            "(436, 1)\n",
            "(338, 1)\n",
            "(291, 1)\n",
            "(198, 0)\n",
            "(18, 0)\n",
            "(334, 1)\n",
            "(215, 0)\n",
            "(67, 1)\n",
            "(485, 1)\n",
            "(247, 1)\n",
            "(76, 1)\n",
            "(352, 0)\n",
            "(550, 1)\n",
            "(308, 1)\n",
            "(5, 0)\n",
            "(286, 1)\n",
            "(499, 0)\n",
            "(517, 0)\n",
            "(165, 1)\n",
            "(358, 1)\n",
            "(217, 1)\n",
            "(181, 0)\n",
            "(151, 1)\n",
            "(155, 1)\n",
            "(474, 1)\n",
            "(330, 0)\n",
            "(229, 0)\n",
            "(552, 1)\n",
            "(496, 1)\n",
            "(212, 0)\n",
            "(238, 1)\n",
            "(447, 1)\n",
            "(72, 0)\n",
            "(270, 1)\n",
            "(375, 1)\n",
            "(230, 0)\n",
            "(404, 1)\n",
            "(322, 1)\n",
            "(187, 1)\n",
            "(195, 1)\n",
            "(156, 0)\n",
            "(91, 0)\n",
            "(46, 1)\n",
            "(244, 0)\n",
            "(90, 1)\n",
            "(316, 1)\n",
            "(475, 1)\n",
            "(234, 1)\n",
            "(294, 1)\n",
            "(143, 1)\n",
            "(47, 0)\n",
            "(24, 0)\n",
            "(534, 1)\n",
            "(154, 1)\n",
            "(116, 1)\n",
            "(258, 0)\n",
            "(233, 0)\n",
            "(171, 0)\n",
            "(424, 1)\n",
            "(331, 1)\n",
            "(383, 1)\n",
            "(221, 1)\n",
            "(170, 1)\n",
            "(520, 1)\n",
            "(117, 0)\n",
            "(335, 0)\n",
            "(4, 0)\n",
            "(110, 1)\n",
            "(52, 1)\n",
            "(206, 1)\n",
            "(318, 1)\n",
            "(95, 0)\n",
            "(536, 0)\n",
            "(415, 1)\n",
            "(379, 0)\n",
            "(342, 1)\n",
            "(391, 1)\n",
            "(451, 0)\n",
            "(118, 0)\n",
            "(241, 1)\n",
            "(406, 1)\n",
            "(459, 1)\n",
            "(254, 0)\n",
            "(372, 0)\n",
            "(210, 0)\n",
            "(175, 1)\n",
            "(518, 1)\n",
            "(257, 0)\n",
            "(298, 1)\n",
            "(235, 1)\n",
            "(507, 1)\n",
            "(546, 1)\n",
            "(366, 0)\n",
            "(160, 1)\n",
            "(167, 0)\n",
            "(478, 1)\n",
            "(416, 1)\n",
            "(163, 1)\n",
            "(137, 1)\n",
            "(443, 1)\n",
            "(243, 1)\n",
            "(87, 0)\n",
            "(450, 1)\n",
            "(323, 0)\n",
            "(462, 1)\n",
            "(376, 1)\n",
            "(157, 1)\n",
            "(133, 1)\n",
            "(93, 1)\n",
            "(225, 1)\n",
            "(250, 0)\n",
            "(223, 0)\n",
            "(328, 0)\n",
            "(408, 0)\n",
            "(245, 1)\n",
            "(515, 1)\n",
            "(348, 1)\n",
            "(231, 1)\n",
            "(281, 1)\n",
            "(568, 1)\n",
            "(519, 1)\n",
            "(42, 0)\n",
            "(412, 1)\n",
            "(367, 1)\n",
            "(59, 1)\n",
            "(265, 0)\n",
            "(205, 0)\n",
            "(82, 0)\n",
            "(107, 1)\n",
            "(494, 1)\n",
            "(511, 1)\n",
            "(387, 1)\n",
            "(553, 1)\n",
            "(504, 1)\n",
            "(138, 0)\n",
            "(513, 1)\n",
            "(385, 0)\n",
            "(471, 1)\n",
            "(448, 1)\n",
            "(69, 1)\n",
            "(14, 0)\n",
            "(40, 0)\n",
            "(321, 0)\n",
            "(189, 1)\n",
            "(542, 1)\n",
            "(251, 1)\n",
            "(62, 0)\n",
            "(558, 1)\n",
            "(538, 1)\n",
            "(350, 1)\n",
            "(353, 0)\n",
            "(422, 1)\n",
            "(239, 0)\n",
            "(340, 1)\n",
            "(66, 1)\n",
            "(493, 1)\n",
            "(269, 1)\n",
            "(482, 1)\n",
            "(99, 0)\n",
            "(219, 0)\n",
            "(271, 1)\n",
            "(481, 1)\n",
            "(458, 1)\n",
            "(119, 0)\n",
            "(389, 0)\n",
            "(164, 0)\n",
            "(539, 1)\n",
            "(417, 0)\n",
            "(491, 1)\n",
            "(470, 1)\n",
            "(7, 0)\n",
            "(505, 1)\n",
            "(130, 1)\n",
            "(319, 1)\n",
            "(337, 0)\n",
            "(199, 0)\n",
            "(113, 1)\n",
            "(433, 0)\n",
            "(81, 1)\n",
            "(194, 0)\n",
            "(176, 1)\n",
            "(202, 0)\n",
            "(437, 1)\n",
            "(371, 1)\n",
            "(214, 0)\n",
            "(541, 1)\n",
            "(554, 1)\n",
            "(341, 1)\n",
            "(441, 0)\n",
            "(17, 0)\n",
            "(407, 1)\n",
            "(6, 0)\n",
            "(296, 1)\n",
            "(20, 1)\n",
            "(369, 0)\n",
            "(431, 1)\n",
            "(51, 1)\n",
            "(418, 1)\n",
            "(55, 1)\n",
            "(142, 1)\n",
            "(200, 1)\n",
            "(564, 0)\n",
            "(150, 1)\n",
            "(242, 1)\n",
            "(185, 1)\n",
            "(141, 0)\n",
            "(421, 1)\n",
            "(43, 0)\n",
            "(359, 1)\n",
            "(456, 1)\n",
            "(98, 1)\n",
            "(373, 0)\n",
            "(236, 0)\n",
            "(263, 0)\n",
            "(525, 1)\n",
            "(102, 1)\n",
            "(140, 1)\n",
            "(120, 1)\n",
            "(127, 0)\n",
            "(31, 0)\n",
            "(267, 1)\n",
            "(439, 1)\n",
            "(60, 1)\n",
            "(446, 0)\n",
            "(311, 1)\n",
            "(204, 1)\n",
            "(325, 1)\n",
            "(315, 1)\n",
            "(356, 1)\n",
            "(521, 0)\n",
            "(382, 1)\n",
            "(89, 1)\n",
            "(249, 1)\n",
            "(354, 1)\n",
            "(275, 1)\n",
            "(317, 0)\n",
            "(396, 1)\n",
            "(262, 0)\n",
            "(208, 1)\n",
            "(390, 1)\n",
            "(304, 1)\n",
            "(174, 1)\n",
            "(196, 0)\n",
            "(104, 1)\n",
            "(96, 1)\n",
            "(559, 1)\n",
            "(522, 1)\n",
            "(54, 0)\n",
            "(232, 1)\n",
            "(136, 1)\n",
            "(543, 1)\n",
            "(78, 0)\n",
            "(1, 0)\n",
            "(203, 0)\n",
            "(377, 1)\n",
            "(394, 1)\n",
            "(149, 1)\n",
            "(32, 0)\n",
            "(144, 1)\n",
            "(545, 1)\n",
            "(565, 0)\n",
            "(293, 1)\n",
            "(530, 1)\n",
            "(426, 1)\n",
            "(73, 0)\n",
            "(3, 0)\n",
            "(411, 1)\n",
            "(362, 1)\n",
            "(30, 0)\n",
            "(129, 0)\n",
            "(378, 1)\n",
            "(487, 0)\n",
            "(413, 1)\n",
            "(148, 1)\n",
            "(213, 0)\n",
            "(159, 1)\n",
            "(454, 1)\n",
            "(240, 1)\n",
            "(329, 0)\n",
            "(344, 1)\n",
            "(302, 0)\n",
            "(191, 1)\n",
            "(216, 1)\n",
            "(19, 1)\n",
            "(279, 1)\n",
            "(498, 0)\n",
            "(255, 0)\n",
            "(336, 1)\n",
            "(381, 1)\n",
            "(510, 1)\n",
            "(540, 1)\n",
            "(100, 0)\n",
            "(29, 0)\n",
            "(320, 1)\n",
            "(22, 0)\n",
            "(97, 1)\n",
            "(310, 1)\n",
            "(531, 1)\n",
            "(397, 1)\n",
            "(94, 0)\n",
            "(12, 0)\n",
            "(455, 1)\n",
            "(237, 0)\n",
            "(75, 0)\n",
            "(556, 1)\n",
            "(442, 1)\n",
            "(139, 1)\n",
            "(112, 1)\n",
            "(10, 0)\n",
            "(560, 1)\n",
            "(26, 0)\n",
            "(177, 0)\n",
            "(8, 0)\n",
            "(548, 1)\n",
            "(268, 1)\n",
            "(34, 0)\n",
            "(123, 1)\n",
            "(486, 1)\n",
            "(207, 0)\n",
            "(357, 1)\n",
            "(409, 1)\n",
            "(211, 1)\n",
            "(289, 1)\n",
            "(11, 0)\n",
            "(53, 0)\n",
            "(86, 0)\n",
            "(266, 1)\n",
            "(398, 1)\n",
            "(77, 0)\n",
            "(125, 1)\n",
            "(537, 1)\n",
            "(532, 1)\n",
            "(472, 1)\n",
            "(259, 0)\n",
            "(227, 1)\n",
            "(476, 1)\n",
            "(453, 1)\n",
            "(33, 0)\n",
            "(65, 0)\n",
            "(188, 1)\n",
            "(479, 0)\n",
            "(509, 0)\n",
            "(201, 0)\n",
            "(183, 1)\n",
            "(285, 1)\n",
            "(49, 1)\n",
            "(161, 0)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "GvALnH8GJ8nI",
        "colab_type": "code",
        "outputId": "cb321125-5335-48a4-a50d-29a85f00278c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 581
        }
      },
      "source": [
        "print(type(row.shape))\n",
        "print()\n",
        "print(type(X_binarised_train), X_binarised_train.shape, type(X_binarised_train.shape))\n",
        "print()\n",
        "print(type(items)) # tuple object\n",
        "print()\n",
        "print(mod_pred_train)\n",
        "print(type(mod_pred_train))\n",
        "print(mod_pred_train[2], type(mod_pred_train[2]))\n",
        "print(mod_pred_train.index(False))\n",
        "print()\n",
        "fy = [False, False, False, True, True, False]\n",
        "ty = [True, True, True, True, False]\n",
        "ny = [99, 25, 541521, \"sdfsfdsf\", False, False, \"False\", True]\n",
        "print(type(ny.index(541521)), ny.index(541521))\n",
        "print(type(ny.index(False)), ny.index(False))\n",
        "print(type(ny.index(\"False\")), ny.index(\"False\"))\n",
        "print()\n",
        "print(ny[2], type(ny[2]))\n",
        "print(ny[3], type(ny[3]))\n",
        "print(ny[4], type(ny[4]))\n",
        "print()\n",
        "#print(ny.index(\"541521\"))\n",
        "print(type(ny.index(\"sdfsfdsf\")), ny.index(\"sdfsfdsf\"))\n",
        "#print(ny.index(sdfsfdsf))\n",
        "print(mod_pred_train.index(True))\n",
        "print(mod_pred_train.index(False))\n",
        "print(ty.index(True))\n",
        "print(fy.index(False))\n",
        "print()\n",
        "\n",
        "''' The above code will return the index of the element.\n",
        "\n",
        "Since we have a list of boolean values in fy the usage of index()\n",
        "\n",
        "returns only the index of the first occurrence of the value of False.\n",
        "\n",
        "So an alternate method to get the indices of false we go for filter, lambda functions. '''\n",
        "\n",
        "# False indices\n",
        "print(fy)\n",
        "print (\"The original list is : \", fy)\n",
        "# lambda + filter() + range()  \n",
        "f_indices = list(filter(lambda i: not fy[i], range(len(fy))))\n",
        "print (\"List of indices with False values: \", f_indices)\n",
        "print()\n",
        "# True indices\n",
        "print(ty)\n",
        "print (\"The original list is : \", ty)\n",
        "# lambda + filter() + range() \n",
        "t_indices = list(filter(lambda i: ty[i], range(len(ty))))\n",
        "print (\"List of indices with True values: \", t_indices)"
      ],
      "execution_count": 246,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "<class 'tuple'>\n",
            "\n",
            "<class 'numpy.ndarray'> (512, 30) <class 'tuple'>\n",
            "\n",
            "<class 'tuple'>\n",
            "\n",
            "[False, True, False, True, True, False, True, False, False, False, False, False, True, True, False, True, False, False, True, True, False, False, True, True, False, False, False, True, True, False, False, False, False, False, False, True, True, False, False, True, False, False, False, True, False, False, True, False, False, True, True, True, True, False, False, True, False, False, True, False, False, False, True, True, False, False, False, True, True, False, True, False, False, True, False, True, False, False, True, False, False, False, False, True, False, False, False, True, False, True, True, False, False, True, False, False, False, False, True, False, True, False, False, False, True, True, False, False, False, True, False, False, False, False, True, False, True, True, True, False, False, True, False, False, True, False, False, False, False, True, True, True, True, True, True, False, False, True, True, True, False, False, False, False, True, False, False, True, True, False, False, False, False, True, True, True, False, True, True, True, False, True, False, False, True, True, False, False, False, False, True, False, True, False, False, False, False, True, True, False, True, False, False, True, True, True, False, False, True, True, False, False, False, True, False, False, True, False, True, True, False, True, False, True, True, False, False, True, False, True, True, True, True, True, True, False, False, True, True, False, False, False, True, False, True, True, True, True, False, False, False, False, True, True, False, False, False, False, True, False, True, False, False, False, True, True, False, False, False, False, True, False, False, True, True, False, True, False, False, False, True, False, True, True, False, True, False, True, False, False, False, False, True, True, True, False, False, False, False, False, True, True, False, True, True, False, False, True, True, False, False, True, False, True, True, True, True, True, False, False, True, False, False, True, True, False, True, False, True, False, True, False, True, False, True, False, False, False, True, False, True, False, True, False, False, True, True, False, False, False, False, False, False, False, False, False, False, True, False, False, False, False, False, False, False, False, False, True, True, False, False, False, True, False, False, True, False, True, True, False, True, True, True, True, False, True, False, False, False, True, True, False, False, True, False, True, False, False, False, False, False, True, False, False, False, True, True, False, False, True, True, True, True, False, False, True, False, True, True, False, False, True, False, False, True, True, True, False, False, True, False, True, False, False, False, False, False, False, False, False, True, True, False, True, True, False, True, True, True, True, False, True, True, False, False, True, False, False, False, False, True, True, True, False, False, False, True, True, True, False, False, False, True, True, True, True, False, False, True, False, False, True, False, True, False, False, False, False, False, False, True, False, False, False, False, False, False, False, True, True, False, False, True, True, True, True, True, True, False, False, False, True, True, False, True, False, True, True, False, True, True, False, False, False, True, False, False, False, True, True, False, False]\n",
            "<class 'list'>\n",
            "False <class 'bool'>\n",
            "0\n",
            "\n",
            "<class 'int'> 2\n",
            "<class 'int'> 4\n",
            "<class 'int'> 6\n",
            "\n",
            "541521 <class 'int'>\n",
            "sdfsfdsf <class 'str'>\n",
            "False <class 'bool'>\n",
            "\n",
            "<class 'int'> 3\n",
            "1\n",
            "0\n",
            "0\n",
            "0\n",
            "\n",
            "[False, False, False, True, True, False]\n",
            "The original list is :  [False, False, False, True, True, False]\n",
            "List of indices with False values:  [0, 1, 2, 5]\n",
            "\n",
            "[True, True, True, True, False]\n",
            "The original list is :  [True, True, True, True, False]\n",
            "List of indices with True values:  [0, 1, 2, 3]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "pWkpdRvYFOBw",
        "colab_type": "text"
      },
      "source": [
        "# MP Neuron and Perceptron using Python: Class"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "s71G7i_YDkeB",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Using Class to templatize the model\n",
        "class MPNeuron:\n",
        "  \n",
        "  # Initialization function\n",
        "  def __init__(self):\n",
        "    # we define the parameters of the model\n",
        "    # single parameter 'b' for MP Neuron\n",
        "    self.b = None\n",
        "  \n",
        "  # Inference function\n",
        "  # Takes the input data and predicts the output  \n",
        "  def model(self, x):\n",
        "    # returns true if 1 or false if 0\n",
        "    return(sum(x) >= self.b)\n",
        "\n",
        "  # Predict function\n",
        "  # Takes entire set of input data rows and predict the corresponding Y output\n",
        "  def predict(self, X):\n",
        "    Y = []\n",
        "    for x in X:\n",
        "      result = self.model(x)\n",
        "      Y.append(result)    \n",
        "    #print(Y)\n",
        "    return np.array(Y)\n",
        "\n",
        "  # Fit function - to find the right value of b from all possible options\n",
        "  # learning algorithm\n",
        "  def fit(self, X, Y):\n",
        "    accuracy = {}\n",
        "    for b in range(X.shape[1] + 1):\n",
        "      self.b = b\n",
        "      Y_pred = self.predict(X)\n",
        "      accuracy[b] = accuracy_score(Y_pred, Y)\n",
        "    best_b = max(accuracy, key = accuracy.get)\n",
        "    self.b = best_b\n",
        "\n",
        "    print(\"Optimal value of b: \", best_b)\n",
        "    print(\"Highest accuracy: \", accuracy[best_b])"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "BxpfD5eKUXo0",
        "colab_type": "code",
        "outputId": "058138f1-5026-416e-b6e7-0f0ffa4dee0e",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        }
      },
      "source": [
        "# How to use the MP Neuron class model\n",
        "mp_neuron = MPNeuron()\n",
        "mp_neuron.fit(X_binarised_train, Y_train)"
      ],
      "execution_count": 248,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Optimal value of b:  28\n",
            "Highest accuracy:  0.849609375\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "WHe0AG6NYDHl",
        "colab_type": "code",
        "outputId": "59a1c342-4b6e-4f19-bcda-1bfb4aa58720",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        }
      },
      "source": [
        "# Testing phase of the MP Neuron Model\n",
        "Y_test_pred = mp_neuron.predict(X_binarised_test)\n",
        "accuracy_test = accuracy_score(Y_test_pred, Y_test)\n",
        "print(accuracy_test)\n",
        "print(type(Y_test_pred))"
      ],
      "execution_count": 249,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "0.7894736842105263\n",
            "<class 'numpy.ndarray'>\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "lUD3GuJwKJIQ",
        "colab_type": "text"
      },
      "source": [
        "# Perceptron model\n",
        "$y = 1, \\mbox{if}\\sum_i w_i x_i >= b$\n",
        "\n",
        "$y = 0, otherwise$"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "NdFd4t-HoQuA",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "class Perceptron:\n",
        "  # initialize parameters\n",
        "  def __init__(self):\n",
        "    # w and b are weights and bias\n",
        "    self.w = None # w is an array vector\n",
        "    self.b = None # b is a scalar\n",
        "\n",
        "  # Model for a single given input (given an x predict y)\n",
        "  # Inference problem\n",
        "  def model(self, x):\n",
        "    return 1 if (np.dot(self.w, x) >= self.b) else 0\n",
        "\n",
        "\n",
        "  # Model for a entire set of input (given X predict Y) or \n",
        "  # takes vector of x as input and predicts vector of Y as output\n",
        "  # Inference problem\n",
        "  def predict(self, X):\n",
        "    Y = [] # vector \n",
        "    for x in X:\n",
        "      result = self.model(x)\n",
        "      Y.append(result) # list of scalars or list vector (slow)\n",
        "    return np.array(Y) # array vector (fast) -> Y predicted\n",
        "\n",
        "  # given X and Y learn the right parameters    \n",
        "  def fit(self, X, Y):\n",
        "    # Model has to deal with all possible combinations of w and b\n",
        "    # Unlike MP Neuron model where 'b' is just few integers (#features + 1), \n",
        "    # whereas in Perceptron model 'b' is any real valued number and so is the \n",
        "    # case with 'w'. Therefore we cannot go for brute force search because it will\n",
        "    # be inefficient.\n",
        "    # Therefore we have learning algorithm for Perceptron model (w = w + x and \n",
        "    # w = w- x)\n",
        "    # Learning problem\n",
        "    # Inital values of w and b\n",
        "    self.w = np.ones(X.shape[1])\n",
        "    self.b = 0\n",
        "    \n",
        "    for x, y in zip(X, Y): # X train and Y train or actual or ground truth\n",
        "      y_pred = self.model(x)\n",
        "      # Learning occurs here (in case of error between actual and predicted y)\n",
        "      if y == 1 and y_pred == 0: # cosine angle is obtuse\n",
        "        # To make cosine angle acute (cosine value within 1 to 0)\n",
        "        self.w = self.w + x\n",
        "        self.b = self.b + 1 # bias can take any real value\n",
        "      elif y == 0 and y_pred == 1: # cosine angle is acute\n",
        "        # To make cosine angle obtuse (cosine value within 0 to -1)\n",
        "        self.w = self.w - x\n",
        "        self.b = self.b - 1 # bias can take any real value"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bh3xnYEKj8Xo",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "perceptron = Perceptron() # perceptron object is an instance of class Perceptron\n",
        "\n",
        "# Run it with original data (NumPy data :D) and not the binarised data\n",
        "# X train and Y train\n",
        "# Converting into NumPy arrays\n",
        "X_train = X_train.values\n",
        "X_test = X_test.values"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "CljCf7HLk5rA",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Training the model (model is learning)\n",
        "perceptron.fit(X_train, Y_train)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "5Cg_XTF2lngy",
        "colab_type": "code",
        "outputId": "814188c5-feda-448b-de24-9ceb348c7589",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 268
        }
      },
      "source": [
        "plt.plot(perceptron.w)\n",
        "plt.show()  "
      ],
      "execution_count": 253,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "n-0mUTV_mAcc",
        "colab_type": "code",
        "outputId": "c90e4f0f-da5b-4de4-8d62-42bc030f2a56",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "# Accuracy of training data set\n",
        "Y_pred_train = perceptron.predict(X_train)\n",
        "print(accuracy_score(Y_pred_train, Y_train)) # Poor"
      ],
      "execution_count": 254,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "0.373046875\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "VR5gDWmFmOjf",
        "colab_type": "code",
        "outputId": "c8896a81-6462-490b-eb47-068aa3206f6d",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "# Accuracy of test data set\n",
        "Y_pred_test = perceptron.predict(X_test)\n",
        "print(accuracy_score(Y_pred_test, Y_test)) # Poor"
      ],
      "execution_count": 255,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "0.3684210526315789\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jGFuyK_GnLKs",
        "colab_type": "code",
        "outputId": "14ed5255-9f89-4b01-e497-f3c7c9149eb6",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 85
        }
      },
      "source": [
        "print(type(X_train))\n",
        "print(type(Y_train))\n",
        "print(type(X_test))\n",
        "print(type(Y_test))"
      ],
      "execution_count": 256,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "<class 'numpy.ndarray'>\n",
            "<class 'pandas.core.series.Series'>\n",
            "<class 'numpy.ndarray'>\n",
            "<class 'pandas.core.series.Series'>\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Kld3ncz2e-bS",
        "colab_type": "text"
      },
      "source": [
        "## Epochs"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "DrTTTsiboRt6",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# How to improve accuracy?\n",
        "# Epoch\n",
        "# Number of epochs (how many times the model has to \n",
        "# go through the dataset) becomes the hyper-parameter which also decides the \n",
        "# accuracy of the model while training (but it is not a parameter used inside the\n",
        "# model)\n",
        "\n",
        "# Most of the DL is about finding the right value of the hyper-parameter also \n",
        "# known as hyper-parameter tuning\n",
        "class Perceptron:\n",
        "  # initialize parameters\n",
        "  def __init__(self):\n",
        "    # w and b are weights and bias\n",
        "    self.w = None # w is an array vector\n",
        "    self.b = None # b is a scalar\n",
        "\n",
        "  # Model for a single given input (given an x predict y)\n",
        "  # Inference problem\n",
        "  def model(self, x):\n",
        "    return 1 if (np.dot(self.w, x) >= self.b) else 0\n",
        "\n",
        "\n",
        "  # Model for a entire set of input (given X predict Y) or \n",
        "  # takes vector of x as input and predicts vector of Y as output\n",
        "  # Inference problem\n",
        "  def predict(self, X):\n",
        "    Y = [] # vector \n",
        "    for x in X:\n",
        "      result = self.model(x)\n",
        "      Y.append(result) # list of scalars or list vector (slow)\n",
        "    return np.array(Y) # array vector (fast) -> Y predicted\n",
        "\n",
        "  # given X and Y learn the right parameters    \n",
        "  def fit(self, X, Y, epochs = 1):\n",
        "    # Model has to deal with all possible combinations of w and b\n",
        "    # Unlike MP Neuron model where 'b' is just few integers (#features + 1), \n",
        "    # whereas in Perceptron model 'b' is any real valued number and so is the \n",
        "    # case with 'w'. Therefore we cannot go for brute force search because it will\n",
        "    # be inefficient.\n",
        "    # Therefore we have learning algorithm for Perceptron model (w = w + x and \n",
        "    # w = w- x)\n",
        "    # Learning problem\n",
        "    # Inital values of w and b\n",
        "    self.w = np.ones(X.shape[1])\n",
        "    self.b = 0\n",
        "    # For plotting training accuracy\n",
        "    accuracy = {} # dictionary\n",
        "    max_accuracy = 0\n",
        "    for i in range(epochs):\n",
        "      for x, y in zip(X, Y): # X train and Y train or actual or ground truth\n",
        "        y_pred = self.model(x)\n",
        "        # Learning occurs here (in case of error between actual and predicted y)\n",
        "        if y == 1 and y_pred == 0: # cosine angle is obtuse\n",
        "          # To make cosine angle acute (cosine value within 1 to 0)\n",
        "          self.w = self.w + x\n",
        "          self.b = self.b + 1 # bias can take any real value\n",
        "        elif y == 0 and y_pred == 1: # cosine angle is acute\n",
        "          # To make cosine angle obtuse (cosine value within 0 to -1)\n",
        "          self.w = self.w - x\n",
        "          self.b = self.b - 1 # bias can take any real value\n",
        "      # computes accuracy after each epoch    \n",
        "      accuracy[i] = accuracy_score(self.predict(X), Y)\n",
        "      if (accuracy[i] > max_accuracy):\n",
        "        max_accuracy = accuracy[i]\n",
        "    print(max_accuracy)    \n",
        "    #plt.plot(accuracy.values())\n",
        "    # To plot input to be passed for accuracy.values() and loss .values() is of \n",
        "    # type list() when using dictionary for accuracy and loss\n",
        "    plt.plot(list(accuracy.values()))\n",
        "    plt.show()"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "D_D1nonBtOgv",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "perceptron = Perceptron() # perceptron object is an instance of class Perceptron"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qSuBbmXW6DZs",
        "colab_type": "code",
        "outputId": "d7884bd2-903e-48eb-b74b-48f6509be03a",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 282
        }
      },
      "source": [
        "# Training the model and accuracy (model is learning)\n",
        "perceptron.fit(X_train, Y_train, 50) # Accuracy oscillates when epochs are changed\n"
      ],
      "execution_count": 259,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "0.921875\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "YJVit9hMfCEQ",
        "colab_type": "text"
      },
      "source": [
        "## Checkpointing (tracking and saving the best model)"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Ea_9mXhp6oVA",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Issues:\n",
        "#1. Oscillation of accuracy \n",
        "#2. Accuracy increases as epoch increases \n",
        "\n",
        "\n",
        "# How to improve accuracy?\n",
        "# Epoch\n",
        "# Number of epochs (how many times the model has to \n",
        "# go through the dataset) becomes the hyper-parameter which also decides the \n",
        "# accuracy of the model while training (but it is not a parameter used inside the\n",
        "# model)\n",
        "\n",
        "# Most of the DL is about finding the right value of the hyper-parameter also \n",
        "# known as hyper-parameter tuning\n",
        "class Perceptron:\n",
        "  # initialize parameters\n",
        "  def __init__(self):\n",
        "    # w and b are weights and bias\n",
        "    self.w = None # w is an array vector\n",
        "    self.b = None # b is a scalar\n",
        "\n",
        "  # Model for a single given input (given an x predict y)\n",
        "  # Inference problem\n",
        "  def model(self, x):\n",
        "    return 1 if (np.dot(self.w, x) >= self.b) else 0\n",
        "\n",
        "\n",
        "  # Model for a entire set of input (given X predict Y) or \n",
        "  # takes vector of x as input and predicts vector of Y as output\n",
        "  # Inference problem\n",
        "  def predict(self, X):\n",
        "    Y = [] # vector \n",
        "    for x in X:\n",
        "      result = self.model(x)\n",
        "      Y.append(result) # list of scalars or list vector (slow)\n",
        "    return np.array(Y) # array vector (fast) -> Y predicted\n",
        "\n",
        "  # given X and Y learn the right parameters    \n",
        "  def fit(self, X, Y, epochs = 1):\n",
        "    # Model has to deal with all possible combinations of w and b\n",
        "    # Unlike MP Neuron model where 'b' is just few integers (#features + 1), \n",
        "    # whereas in Perceptron model 'b' is any real valued number and so is the \n",
        "    # case with 'w'. Therefore we cannot go for brute force search because it will\n",
        "    # be inefficient.\n",
        "    # Therefore we have learning algorithm for Perceptron model (w = w + x and \n",
        "    # w = w- x)\n",
        "    # Learning problem\n",
        "    # Inital values of w and b\n",
        "    self.w = np.ones(X.shape[1])\n",
        "    self.b = 0\n",
        "    # For plotting training accuracy\n",
        "    accuracy = {} # dictionary\n",
        "    max_accuracy = 0\n",
        "    for i in range(epochs):\n",
        "      for x, y in zip(X, Y): # X train and Y train or actual or ground truth\n",
        "        y_pred = self.model(x)\n",
        "        # Learning occurs here (in case of error between actual and predicted y)\n",
        "        if y == 1 and y_pred == 0: # cosine angle is obtuse\n",
        "          # To make cosine angle acute (cosine value within 1 to 0)\n",
        "          self.w = self.w + x\n",
        "          self.b = self.b + 1 # bias can take any real value\n",
        "        elif y == 0 and y_pred == 1: # cosine angle is acute\n",
        "          # To make cosine angle obtuse (cosine value within 0 to -1)\n",
        "          self.w = self.w - x\n",
        "          self.b = self.b - 1 # bias can take any real value\n",
        "      # computes accuracy after each epoch    \n",
        "      accuracy[i] = accuracy_score(self.predict(X), Y)\n",
        "      #print(accuracy[i])\n",
        "      if (accuracy[i] > max_accuracy):\n",
        "        max_accuracy = accuracy[i]\n",
        "        # Whenever there is an increase in accuracy we store those parameter values\n",
        "        chkptw = self.w\n",
        "        chkptb = self.b \n",
        "    # When model ends at a particular point we ensure that it ends up where\n",
        "    # model parameters has the highest training accuracy (Remember we do not\n",
        "    # have control over test data i.e., we are only maximizing the training accuracy)    \n",
        "    self.w = chkptw\n",
        "    self.b = chkptb\n",
        "    print(max_accuracy)    \n",
        "    #plt.plot(accuracy.values())\n",
        "    # To plot input to be passed for accuracy.values() and loss .values() is of \n",
        "    # type list() when using dictionary for accuracy and loss\n",
        "    plt.plot(list(accuracy.values()))\n",
        "    plt.show()\n",
        "    print(type(chkptw))\n",
        "    print(type(chkptb))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "8yDbnxAyfEwb",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 333
        },
        "outputId": "e5796c75-6d98-4237-df07-db05ea0118fa"
      },
      "source": [
        "perceptron = Perceptron() # perceptron object is an instance of class Perceptron\n",
        "# Training the model and accuracy (model is learning)\n",
        "\n",
        "# We can see that the model runs with the parameters that gives the maximum accuracy seen\n",
        "# as we have used checkpoininting that saves the weights and bias of maximum accuracy \n",
        "# after running through different epochs across the dataset\n",
        "\n",
        "# Checkpoiting keeps track of best model seen and saving that model\n",
        "perceptron.fit(X_train, Y_train, 100) # Accuracy oscillates when epochs are changed\n",
        "# Accuracy of training data set\n",
        "\n",
        "Y_pred_train = perceptron.predict(X_train)\n",
        "print(accuracy_score(Y_pred_train, Y_train))"
      ],
      "execution_count": 261,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "0.927734375\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "text": [
            "<class 'numpy.ndarray'>\n",
            "<class 'int'>\n",
            "0.927734375\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Yz0iCaWjaxDs",
        "colab_type": "text"
      },
      "source": [
        "## Learning rate"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Z95-ET6eam9K",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# To smoothen the graph between epoch ans accuracy is through learning rate\n",
        "class Perceptron:\n",
        "  # initialize parameters\n",
        "  def __init__(self):\n",
        "    # w and b are weights and bias\n",
        "    self.w = None # w is an array vector\n",
        "    self.b = None # b is a scalar\n",
        "\n",
        "  # Model for a single given input (given an x predict y)\n",
        "  # Inference problem\n",
        "  def model(self, x):\n",
        "    return 1 if (np.dot(self.w, x) >= self.b) else 0\n",
        "\n",
        "\n",
        "  # Model for a entire set of input (given X predict Y) or \n",
        "  # takes vector of x as input and predicts vector of Y as output\n",
        "  # Inference problem\n",
        "  def predict(self, X):\n",
        "    Y = [] # vector \n",
        "    for x in X:\n",
        "      result = self.model(x)\n",
        "      Y.append(result) # list of scalars or list vector (slow)\n",
        "    return np.array(Y) # array vector (fast) -> Y predicted\n",
        "\n",
        "  # given X and Y learn the right parameters    \n",
        "  def fit(self, X, Y, epochs = 1, lr = 1):\n",
        "    # Model has to deal with all possible combinations of w and b\n",
        "    # Unlike MP Neuron model where 'b' is just few integers (#features + 1), \n",
        "    # whereas in Perceptron model 'b' is any real valued number and so is the \n",
        "    # case with 'w'. Therefore we cannot go for brute force search because it will\n",
        "    # be inefficient.\n",
        "    # Therefore we have learning algorithm for Perceptron model (w = w + x and \n",
        "    # w = w- x)\n",
        "    # Learning problem\n",
        "    # Inital values of w and b\n",
        "    self.w = np.ones(X.shape[1])\n",
        "    self.b = 0\n",
        "    # For plotting training accuracy\n",
        "    accuracy = {} # dictionary\n",
        "    max_accuracy = 0\n",
        "    for i in range(epochs):\n",
        "      for x, y in zip(X, Y): # X train and Y train or actual or ground truth\n",
        "        y_pred = self.model(x)\n",
        "        # Learning occurs here (in case of error between actual and predicted y)\n",
        "        if y == 1 and y_pred == 0: # cosine angle is obtuse\n",
        "          # To make cosine angle acute (cosine value within 1 to 0)\n",
        "          ''' learning happens in these statments (changing w and b by learning\n",
        "          from data) and we like to reduce learning rate in order to improve\n",
        "          the accuracy as well as learning pattern'''\n",
        "          # learning rate - lr is a fraction less than 1\n",
        "          self.w = self.w + lr * x \n",
        "          self.b = self.b + lr * 1 # bias can take any real value\n",
        "        elif y == 0 and y_pred == 1: # cosine angle is acute\n",
        "          # To make cosine angle obtuse (cosine value within 0 to -1)\n",
        "          ''' learning happens in these statments (changing w and b by learning\n",
        "          from data) and we like to reduce learning rate in order to improve\n",
        "          the accuracy as well as learning pattern'''\n",
        "          # learning rate - lr is a fraction less than 1\n",
        "          self.w = self.w - lr * x\n",
        "          self.b = self.b - lr * 1 # bias can take any real value\n",
        "\n",
        "\n",
        "      # computes accuracy after each epoch    \n",
        "      accuracy[i] = accuracy_score(self.predict(X), Y)\n",
        "      #print(accuracy[i])\n",
        "      if (accuracy[i] > max_accuracy):\n",
        "        max_accuracy = accuracy[i]\n",
        "        # Whenever there is an increase in accuracy we store those parameter values\n",
        "        chkptw = self.w\n",
        "        chkptb = self.b \n",
        "    # When model ends at a particular point we ensure that it ends up where\n",
        "    # model parameters has the highest training accuracy (Remember we do not\n",
        "    # have control over test data i.e., we are only maximizing the training accuracy)    \n",
        "    self.w = chkptw\n",
        "    self.b = chkptb\n",
        "    print(max_accuracy)    \n",
        "    #plt.plot(accuracy.values())\n",
        "    # To plot input to be passed for accuracy.values() and loss .values() is of \n",
        "    # type list() when using dictionary for accuracy and loss\n",
        "    plt.plot(list(accuracy.values()))\n",
        "    plt.ylim([0, 1]) # accuracy number setting as 0 to 1\n",
        "    plt.show()\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "eJLPXojrbMbJ",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 303
        },
        "outputId": "4fb3fb33-09f4-4bfa-9747-6ece63d9a0ab"
      },
      "source": [
        "perceptron = Perceptron() # perceptron object is an instance of class Perceptron\n",
        "# Training the model and accuracy (model is learning)\n",
        "\n",
        "# We can see that the model runs with the parameters that gives the maximum accuracy seen\n",
        "# as we have used checkpoininting that saves the weights and bias of maximum accuracy \n",
        "# after running through different epochs across the dataset\n",
        "\n",
        "# Checkpoiting keeps track of best model seen and saving that model\n",
        "# Learning rate with a value less than 1 is passed as an argument to reduce the\n",
        "# the learning rate\n",
        "\n",
        "''' Oscillation seems improved after introducing learning rate which has a constant\n",
        "value below 1 and also improves the accuracy as we tune it diligently \n",
        "i.e., Learning rate is another hyperparameter along with epoch that helps in\n",
        "gradually increasing the accuracy of the model'''\n",
        "\n",
        "# We have two hyperparameters: epoch and learning rate for improving the \n",
        "# accuracy of the model along with the learning parameters associated with the\n",
        "# dataset such as weights and bias\n",
        "\n",
        "''' What we can infer from epoch and learning rate is that we need to tune these\n",
        "hyper-parameters such that we can extract the maximum performance of the model '''\n",
        "\n",
        "# epoch and learning rate both work in conjunction\n",
        "perceptron.fit(X_train, Y_train, 100000, 0.001)\n",
        "# Accuracy of training data set\n",
        "Y_pred_train = perceptron.predict(X_train)\n",
        "\n",
        "# Perceptron model takes real valued inputs and also it has increased the\n",
        "# complexity of the model so as to maximise the performance of the model that\n",
        "# can deal with over all dataset that takes in real inputs (rather than binarising data) \n",
        "# as well as real values for weights and bias \n",
        "print(accuracy_score(Y_pred_train, Y_train))"
      ],
      "execution_count": 263,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "0.958984375\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "text": [
            "0.958984375\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "2r2v8uSBiYnH",
        "colab_type": "text"
      },
      "source": [
        "## Test Accuracy: Let us see how our trained model performs on the test data"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "RB8JqzjHe7Cs",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "34068565-d549-4b3e-b480-660f8e2017dd"
      },
      "source": [
        "Y_pred_test = perceptron.predict(X_test)\n",
        "print(accuracy_score(Y_pred_test, Y_test))\n"
      ],
      "execution_count": 264,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "0.9298245614035088\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bhb-4n7si3oA",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 268
        },
        "outputId": "7e0a978e-464d-401b-c04b-29ceb6964a16"
      },
      "source": [
        "# Plot the weights\n",
        "''' Different features are being weighted by different values by the algorithm '''\n",
        "# features vs weights\n",
        "plt.plot(perceptron.w)\n",
        "plt.show() "
      ],
      "execution_count": 265,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "M3JBXlGgoBAw",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "class Perceptron:\n",
        "  # initialize parameters\n",
        "  def __init__(self):\n",
        "    # w and b are weights and bias\n",
        "    self.w = None # w is an array vector\n",
        "    self.b = None # b is a scalar\n",
        "\n",
        "  # Model for a single given input (given an x predict y)\n",
        "  # Inference problem\n",
        "  def model(self, x):\n",
        "    return 1 if (np.dot(self.w, x) >= self.b) else 0\n",
        "\n",
        "\n",
        "  # Model for a entire set of input (given X predict Y) or \n",
        "  # takes vector of x as input and predicts vector of Y as output\n",
        "  # Inference problem\n",
        "  def predict(self, X):\n",
        "    Y = [] # vector \n",
        "    for x in X:\n",
        "      result = self.model(x)\n",
        "      Y.append(result) # list of scalars or list vector (slow)\n",
        "    return np.array(Y) # array vector (fast) -> Y predicted\n",
        "\n",
        "  # given X and Y learn the right parameters    \n",
        "  def fit(self, X, Y, epochs = 1, lr = 1):\n",
        "    # Model has to deal with all possible combinations of w and b\n",
        "    # Unlike MP Neuron model where 'b' is just few integers (#features + 1), \n",
        "    # whereas in Perceptron model 'b' is any real valued number and so is the \n",
        "    # case with 'w'. Therefore we cannot go for brute force search because it will\n",
        "    # be inefficient.\n",
        "    # Therefore we have learning algorithm for Perceptron model (w = w + x and \n",
        "    # w = w- x)\n",
        "    # Learning problem\n",
        "    # Inital values of w and b\n",
        "    self.w = np.ones(X.shape[1])\n",
        "    self.b = 0\n",
        "    # For plotting training accuracy\n",
        "    accuracy = {} # dictionary\n",
        "    max_accuracy = 0\n",
        "\n",
        "    wt_matrix = [] # visualize purpose of weights and features\n",
        "\n",
        "    for i in range(epochs):\n",
        "      for x, y in zip(X, Y): # X train and Y train or actual or ground truth\n",
        "        y_pred = self.model(x)\n",
        "        # Learning occurs here (in case of error between actual and predicted y)\n",
        "        if y == 1 and y_pred == 0: # cosine angle is obtuse\n",
        "          # To make cosine angle acute (cosine value within 1 to 0)\n",
        "          ''' learning happens in these statments (changing w and b by learning\n",
        "          from data) and we like to reduce learning rate in order to improve\n",
        "          the accuracy as well as learning pattern'''\n",
        "          # learning rate - lr is a fraction less than 1\n",
        "          self.w = self.w + lr * x \n",
        "          self.b = self.b + lr * 1 # bias can take any real value\n",
        "        elif y == 0 and y_pred == 1: # cosine angle is acute\n",
        "          # To make cosine angle obtuse (cosine value within 0 to -1)\n",
        "          ''' learning happens in these statments (changing w and b by learning\n",
        "          from data) and we like to reduce learning rate in order to improve\n",
        "          the accuracy as well as learning pattern'''\n",
        "          # learning rate - lr is a fraction less than 1\n",
        "          self.w = self.w - lr * x\n",
        "          self.b = self.b - lr * 1 # bias can take any real value\n",
        "      wt_matrix.append(self.w) # list that holds weights after each epoch\n",
        "      #print(wt_matrix)\n",
        "\n",
        "      # computes accuracy after each epoch    \n",
        "      accuracy[i] = accuracy_score(self.predict(X), Y)\n",
        "      #print(accuracy[i])\n",
        "      if (accuracy[i] > max_accuracy):\n",
        "        max_accuracy = accuracy[i]\n",
        "        # Whenever there is an increase in accuracy we store those parameter values\n",
        "        chkptw = self.w\n",
        "        chkptb = self.b \n",
        "    # When model ends at a particular point we ensure that it ends up where\n",
        "    # model parameters has the highest training accuracy (Remember we do not\n",
        "    # have control over test data i.e., we are only maximizing the training accuracy)    \n",
        "    self.w = chkptw\n",
        "    self.b = chkptb\n",
        "    print(max_accuracy)    \n",
        "    #plt.plot(accuracy.values())\n",
        "    # To plot input to be passed for accuracy.values() and loss .values() is of \n",
        "    # type list() when using dictionary for accuracy and loss\n",
        "    plt.plot(list(accuracy.values()))\n",
        "    plt.ylim([0, 1]) # accuracy number setting as 0 to 1\n",
        "    plt.show()\n",
        "    #print(wt_matrix)\n",
        "    return np.array(wt_matrix)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "mLMjABbvoi4I",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 286
        },
        "outputId": "c033c28e-52d2-4041-deee-fcc9db032dbe"
      },
      "source": [
        "perceptron = Perceptron()\n",
        "wt_matrix = perceptron.fit(X_train, Y_train, 100)"
      ],
      "execution_count": 267,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "0.927734375\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "46mXsyZro_64",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 476
        },
        "outputId": "7f2c3b6b-7c9c-41d3-e2d0-0bd24de753ee"
      },
      "source": [
        "print(type(wt_matrix))\n",
        "print(wt_matrix.shape) # 100 epochs and 30 features\n",
        "print(wt_matrix.size)\n",
        "print(wt_matrix.ndim)\n",
        "print(type(wt_matrix[0]))\n",
        "print(wt_matrix[0].ndim)\n",
        "# prints first row which contains the weights associated to \n",
        "# 30 features of first epoch\n",
        "print(wt_matrix[0])\n",
        "# prints the weights of 100 epochs as a matrix\n",
        "print(wt_matrix)"
      ],
      "execution_count": 268,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "<class 'numpy.ndarray'>\n",
            "(100, 30)\n",
            "3000\n",
            "2\n",
            "<class 'numpy.ndarray'>\n",
            "1\n",
            "[ 4.4787000e+02  8.3489000e+02  2.7157700e+03  2.0408000e+03\n",
            "  5.7084700e+00  2.0157400e+00 -2.3629660e+00 -7.8631100e-01\n",
            "  9.9227000e+00  4.5390700e+00  3.2492000e+00  6.7091400e+01\n",
            "  1.3170500e+01 -1.1429410e+03  1.3688870e+00  1.4125040e+00\n",
            "  1.2378939e+00  1.1849880e+00  2.0698010e+00  1.1560210e+00\n",
            "  4.4626900e+02  1.0824500e+03  2.6876200e+03 -3.7992000e+03\n",
            "  7.2313300e+00  1.2725100e+00 -3.1415070e+00 -7.8600000e-02\n",
            "  1.4028700e+01  4.9343900e+00]\n",
            "[[ 4.4787000e+02  8.3489000e+02  2.7157700e+03 ... -7.8600000e-02\n",
            "   1.4028700e+01  4.9343900e+00]\n",
            " [ 7.2467000e+02  1.3272100e+03  4.3620200e+03 ... -1.6086000e+00\n",
            "   2.1280100e+01  7.0385800e+00]\n",
            " [ 8.9946300e+02  1.6385200e+03  5.3786600e+03 ... -3.3919000e+00\n",
            "   2.5183700e+01  8.1632900e+00]\n",
            " ...\n",
            " [ 4.9431740e+03  9.9315000e+02  2.4199580e+04 ... -2.0482685e+02\n",
            "  -5.1305400e+01 -2.2573060e+01]\n",
            " [ 4.9654040e+03  9.5290000e+02  2.4264710e+04 ... -2.0737302e+02\n",
            "  -5.3126800e+01 -2.3258790e+01]\n",
            " [ 4.9880040e+03  9.4072000e+02  2.4340970e+04 ... -2.0958589e+02\n",
            "  -5.4811100e+01 -2.3852280e+01]]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "MJrWhJdVvXRs",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 265
        },
        "outputId": "f6fbd3f2-3adf-41bd-9da3-5e82977aaf78"
      },
      "source": [
        "# Plot the weights\n",
        "''' Different features are being weighted by different values by the algorithm '''\n",
        "# features vs weights\n",
        "# we can understand what is the range of weights to set the values of y axis for animation\n",
        "plt.plot(wt_matrix[-1,:]) # plots the last row or the 100th epoch \n",
        "plt.show()\n",
        "# see the weights go in the range between 25000 and -10000"
      ],
      "execution_count": 269,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "glJ9ZH58jGuY",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "''' We shall see what all the features play significant role in weight association\n",
        "in the dataset through animation when DL model is used '''\n",
        "\n",
        "%matplotlib inline\n",
        "#import numpy as np\n",
        "#import matplotlib.pyplot as plt\n",
        "\n",
        "from matplotlib import animation, rc\n",
        "from IPython.display import HTML\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "BwKlMh7GkzDP",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 269
        },
        "outputId": "8a6428ea-1cde-4a8e-a6eb-6b798778ff61"
      },
      "source": [
        "# First set up the figure, the axis, and the plot element we want to animate\n",
        "fig, ax = plt.subplots()\n",
        "\n",
        "ax.set_xlim(( 0, wt_matrix.shape[1]))\n",
        "#ax.set_ylim((-10000, 25000))\n",
        "ax.set_ylim((-15000, 25000))\n",
        "\n",
        "line, = ax.plot([], [], lw=2)"
      ],
      "execution_count": 271,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "dvx83YVHk_6a",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# animation function. This is called sequentially\n",
        "def animate(i):\n",
        "    x = list(range(wt_matrix.shape[1])) # the number of features\n",
        "    y = wt_matrix[i, :] # i th row all possible values\n",
        "    line.set_data(x, y)\n",
        "    return (line,)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "g6zvABMzlRk5",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# call the animator. blit=True means only re-draw the parts that have changed.\n",
        "anim = animation.FuncAnimation(fig, animate, frames=100, interval=200, blit=True)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "03dMg-HmlUqr",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 309
        },
        "outputId": "88034aec-c77c-485f-db99-ed665281ecd4"
      },
      "source": [
        "# Training data\n",
        "# 100 epochs = rows = frames with interval 200\n",
        "# Visualizing data iteratively to understand what is happening between the weights and the features\n",
        "HTML(anim.to_html5_video())\n",
        "# As epoch increase the magnitude of weights increase (as complexity of the model increase weight increases may be positive \n",
        "# or negative value which leads to overfitting)"
      ],
      "execution_count": 274,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<video width=\"432\" height=\"288\" controls autoplay loop>\n",
              "  <source type=\"video/mp4\" src=\"data:video/mp4;base64,AAAAHGZ0eXBNNFYgAAACAGlzb21pc28yYXZjMQAAAAhmcmVlAAB/6m1kYXQAAAKtBgX//6ncRem9\n",
              "5tlIt5Ys2CDZI+7veDI2NCAtIGNvcmUgMTUyIHIyODU0IGU5YTU5MDMgLSBILjI2NC9NUEVHLTQg\n",
              "QVZDIGNvZGVjIC0gQ29weWxlZnQgMjAwMy0yMDE3IC0gaHR0cDovL3d3dy52aWRlb2xhbi5vcmcv\n",
              "eDI2NC5odG1sIC0gb3B0aW9uczogY2FiYWM9MSByZWY9MyBkZWJsb2NrPTE6MDowIGFuYWx5c2U9\n",
              "MHgzOjB4MTEzIG1lPWhleCBzdWJtZT03IHBzeT0xIHBzeV9yZD0xLjAwOjAuMDAgbWl4ZWRfcmVm\n",
              "PTEgbWVfcmFuZ2U9MTYgY2hyb21hX21lPTEgdHJlbGxpcz0xIDh4OGRjdD0xIGNxbT0wIGRlYWR6\n",
              "b25lPTIxLDExIGZhc3RfcHNraXA9MSBjaHJvbWFfcXBfb2Zmc2V0PS0yIHRocmVhZHM9MyBsb29r\n",
              "YWhlYWRfdGhyZWFkcz0xIHNsaWNlZF90aHJlYWRzPTAgbnI9MCBkZWNpbWF0ZT0xIGludGVybGFj\n",
              "ZWQ9MCBibHVyYXlfY29tcGF0PTAgY29uc3RyYWluZWRfaW50cmE9MCBiZnJhbWVzPTMgYl9weXJh\n",
              "bWlkPTIgYl9hZGFwdD0xIGJfYmlhcz0wIGRpcmVjdD0xIHdlaWdodGI9MSBvcGVuX2dvcD0wIHdl\n",
              "aWdodHA9MiBrZXlpbnQ9MjUwIGtleWludF9taW49NSBzY2VuZWN1dD00MCBpbnRyYV9yZWZyZXNo\n",
              "PTAgcmNfbG9va2FoZWFkPTQwIHJjPWNyZiBtYnRyZWU9MSBjcmY9MjMuMCBxY29tcD0wLjYwIHFw\n",
              "bWluPTAgcXBtYXg9NjkgcXBzdGVwPTQgaXBfcmF0aW89MS40MCBhcT0xOjEuMDAAgAAAEltliIQA\n",
              "Ev/+963fgU3AQO1rulc4tMurlDQ9UfaUpni2SAAAAwADlGX8OsidmA88y6GZ/02b2g1gjvLH4xTy\n",
              "2fmUGzfV9ZvyIPYmqrCcmekF02fBENtN882c8j8QE2tyBs1WnoR/slNLhn5TRelWS/9cLLoFcr7H\n",
              "HJ7V73RnK6L+vkur0JyIhZgaXyn/gq2d+r2IkXF/g+83w+AMa+bWulIQAAAYByXzwH/pSc3EK3DJ\n",
              "HrSjkvXOhRXq9XPKf8BRMbnZ+5aMyhKsdTA8008IUQsnFZBA5VnZ+azPkbt3VqPcE88+IAeDC0Ql\n",
              "1zYi+Y4PQdsPYL3USLhQBDI2eLfTL7SjsHVO0jcEguKgQFQ3Lfbo+rSx7e88L6cQHv/0GhxAmn0G\n",
              "ZKPyJfVg+xKtqSLidd/vU6faKI4d56jR0yMDhixrUZJfJwjSBn/AYhk+sbowmpDkaVbH8QKkrtp9\n",
              "lt4imkjEQ9HX8R5TEBBwirbj8CSsWugGUbys0uTNNBh+d81pQq+pWjUGvHAJZEfzK2J/G8B6HE3I\n",
              "cyZngvU6mStBsImOeIcksqQuv5ao07qY4Llmxk1pgYFAAFUBNihPSOaxTmQccgGtOe74j57bmQc+\n",
              "HZR2lmrNLvYz337Ns/RBoR4NaOuLxMudvX4H07YBF9lDd+TBQ3cdc5TYoxhoFGsOOlOMJ2prXqog\n",
              "wSmOIYcQY+6k/xzX/BTo/T/Ovufd2329GTSfYXzMf427JvVx2tXHbX3h8C0oPFsa8Z+zl+KaLECv\n",
              "zC87H1LIHBkX9WtKT9DPUMAb3Qf177/YAAbqi3X9OdQnaWA5AfKhhSBpB6vkU6XqrlapL55qVqDj\n",
              "jT5VU6y2t8mtItW2CCbBaiw2vcywYa59cd/aABZAsnD8pEmHcs/YVDWAh2I1AACBtZe2oJD80b16\n",
              "TBpWSCjcFuU3eANnkh+cAz+CvyaPc9yjKWOEGYu+tiLCqnDZEXMoKDjlBZ93KxSxKugKSsWj5sbO\n",
              "oesbjDs2XZVTRfvm1sd8/QBnEeA3MkBx01XeWB/7nwMbdOAbIptFb6Xn0Jz/2fOlAQuXgQ2QA9DB\n",
              "HfipWSH77RYMWkhkFMqaRGOrIllkS+j3Na4VeNmp7uBLQ+eVqPUt39L4aQYfQwiGdus56wl9dRfk\n",
              "39h8H+AKtpxXJQTo9gJRiQ7huISb/OA88l/agi3mJuNRYIOA5FzFT249criAdK6Lt4bDE7z49/e1\n",
              "PQPQYRmp3NOq1SVu2xGBiI+fv5QWkyrUM1YV1eaG8KHFc0BJa1ukrrZgXX4mfEFk8uJvBtvJi9ug\n",
              "4XzqbFl7Zq9pYYK6yAzGfDsCjrvJN+GnvHyAh4z+QzhdFtt7fVD1zvhTEsT3zoUV56QJ/55GZG6N\n",
              "htZlLGbg+behOxiuMVtSfF2nNXzRZjgtw8gh8LmOmI7Z4mFQgQZJ124Gqqk3rsJgQW3+nhLqRE3w\n",
              "4l/KHI7VoAIhOJvTKoLhZHwwtogQQUdevF4J8wsd6SmckOyoxtUc5l9cdJ8aAPjawYWANMcJvvdo\n",
              "2wkxCdFLZRwH5vlsaC5j8ScKmaqi5F+gDi4H8ZrNo8SZfWxbDj32axvYS1SwWjJ4i7tAxoW/UmUG\n",
              "EP4OwfYgAAAGtOKswhT5bDm2szbLIKbSCQkaVcHaj3sZCIzWOewGXAtzUggifdj/HDphk3npHxTm\n",
              "/UZXR4RrUzpDtV0FVgmXjdlkJEfLmRc6EauwibkDAAAO3C9OXJLdAiPc6rhSZOJ2J1sVDO0pBO1N\n",
              "XhQzy+CJ03X0Pp198Wjv7ime/bZ5hYfyGrk6TVf5n1/2wKs7eEtEOzsnNRH9zikEvQ+E4u0SWy/H\n",
              "2qs0Q+lon2Bjh5zaK//eEyUcsxUbeD0gGjCzUHxUIaLUO7YNMCquRhO6AOVUxiwsactoeBYt4M1p\n",
              "jivdp2M3zZbW+ohcKyrtghdOQITUCo7VkHRWh5f5q5LZcHa85xj7bwQ8ntOn3oYDZLRNuk1w91XT\n",
              "mWwOqtibEhbarnfLpi9hlt2vUSDsDSTXTfSNMxVk4j+tovPcZYHw3gU7uPGeDCPFEEM6SO8HSrGP\n",
              "qk3rL3BeFnFjtbL2c/O/jXnINz8Io8LkNS4nMBgqt0s2bl9goqFdCqcVxAqGeenW8U7mgub/s/gM\n",
              "mh1ETreHWTnh+uJWQrSARo+KBq+EzqAF/+G+xJEF5sLo8M1o+U8PHOuHXiCVogUeRdSAeHLFT799\n",
              "YcmKRAOm+sHui/pRmn95u3dwCTt4z76Ap6XAkXjeR/sjme8ZuLJrN7PoftmlKGwzNFPf2usUWf/I\n",
              "R1zjblu08y82vTFzap+eGcbuYwx7QKLo05O3Cnta0hq9gyWjv9rxvhHvyS8nJgjKXWwX6Id6cDYI\n",
              "6G7syzzFgobs8IW5ep+CT3fyMeVE1JQQgJ33/TW/rImvn9fIX8b/nlQM7UQqJQeivtkb+SqmR2Kx\n",
              "kKKqdMStzVrFT/iLA1Ld+IBMGWLl8vHjzicnY4hC67ryZryV/i2jrXLtBnzmEKgjuOdNAZFkWnj1\n",
              "ZWZyKxu+IjbzSAiNyE/Op2aJPZp2/bY/GZyw6fBOkT5eCrnLx200kT3/qIbjKNfRQuH5q9X2z69R\n",
              "etiFUiyLXA6qdGs2m/Tv/+EozIKx/JdQHZXHYs8I6slzMUDnQ6vL0TXuSX5Fy1XBciVi70KoekyA\n",
              "LGJuaexnlZBuSUXkTYUgG/TzWrdzgJ8ulFaFZST67yhfY7y29QhEYGDjsTNxcYphTtdl83wQTtlJ\n",
              "MpZZMQf/9xkFzibgD0gvfmjy5W8va6MMVNYS7kTxa8d5yFDr7K3Wp+2qvrWT7B6j6VYz0tP3MEs/\n",
              "ePvmOFQ2f/K7vPaoYH3rTrY1dPqx4ob0Y0JgB+CaUAVMnvh5hA1leTMvIN7zRFuMKEA3C6/P69LZ\n",
              "3Ppefpa0bQtY29TioNCjtjFCdBHPhDOSnF6o01gTzgGBRVDxEmthIxF1KBwf8IDvDybfeWZQpVI4\n",
              "frI3dTPSBxG4wM+eQ2B6veoYE17aS5GII84rNp1Y3cRhgqwmj62vRqQidj/yS/DHhN7MzhZ5zJ2n\n",
              "ldR867J9dI8ewCnhSKat/FOMk5tP34dWGW+NLbqg5p3ZPYH/9mJPtscjPEpbOtpokZDV6y9iS7Al\n",
              "yDUVWteQvsWe6pZVrP352YhVuiTdydyPs50bkp5xDD+JI4/iXYyZZ+WDmxfHjOn/yLvQXj0hMo6h\n",
              "t+5nk/vnPeRMaNQWICfje4+ezKd1c66w784EXI8CKhNP2ATHX3QS0Rn8aL+BpPs98yijC3m7l1zG\n",
              "oAiTcxhvartxGzH/aeGOHhGPCtXrNdo4i8H/FadzW0Nj3BbpAxcdd8kPMZ0ANqLd0u8k+5Iuv353\n",
              "1SSvA3JncXtNacUNQgFvB/tTbsrej02cu/eeG1aoINlN3OIG9EuDQe/PhpbDqyD5kTlaNUdEGq7k\n",
              "Nx7Pjq3NB0px6B6JcdKQFjd3q3lMhPzxRwFEjmbJuCc+g+oPjaP+quyipCDKVb/kzbWUj6Vn227e\n",
              "ik+noKR7k2NpUOwygZpLCy/tKOgwc2gG3QtByZU07rIFvxwve6uysZNUhR+sOsFRHACMQEtTHVY/\n",
              "YsYS01LsKfSldrDp3lIu6PUdBxkI6aDU0+yK8q3N7x8F2gLPMR1g7qyQhVxp5t0G7lOnDqbKgyYs\n",
              "CsndXk32h5SdCed1ffmOY+KQF14Lc4IG4ToDcXY8HZREsSIsl+mymMgKqgyoueZtSNFABWVvAxrY\n",
              "RD4aseiOLM8KU51Q1QsACpZc7+ax+Ep//ZxDPaX1uzKiF5T46kbWZtmx3o3OPltt3f6tQLVFS0DY\n",
              "CORQE0HNGwolHERUylJzJqJ93nyRWNoRPRATmGV79mUmfnyHeFD8klQN+CszfhvvQUdCzKSyovXC\n",
              "wgvd02tHMxdchcvwL1Pp2Gk6D2vSNmuVYKKFAgiI9eyO6YbHoJxSEbzSQHZYXOZtPaEVrHE8FftO\n",
              "tI6+mNwSix5SKTs6xDBXT1ERFULg4KS1nAz3znZ+pm/jyy+5Zm2tt/Z76PFTLx/4yeVyzztVYwf0\n",
              "N/RiRPgM6ruoSb1FZCL8TWt9GHS3YN6HIXCJFw+TecjICg3V34mO9HDkDGSMmmv9/GcO7ZAOB4sW\n",
              "ISFpL+gV0wCbuFF+Gl+9RdOsEtL5xEYvXcEJ8ztwAwzNQiq/m9ykKsbPjPF8Kq6gErFkJQf026C8\n",
              "anPl8fVZrWwC2zMtnW1Hcxe5kE78L2OgVF3zM35F0QtFIKABDHUSafeIW9xNHwZqvg9SJId0NN6M\n",
              "o/6sZCSHQRLyytiwYHR/SvEzyK8a7LrphhqcXWkICrtt/OzV56h6vWgr1qk8cOhT4uc0yp7w8vfn\n",
              "gJP9CGLhE2ldzqg1xltjfT1o6Wo3BEdBYkcKvc7qrRFvTqKz/MxGbRqGf33r1y10S201xYZaAD54\n",
              "of2mawf8kgyZnX567z/pKR6ZCV05CCmF+DbDpN0LD5V4cyNIHSCviVs9o3SjZY4gDImokZ8pUoX7\n",
              "7xT97QZqcvcqVukH0kycTbrisBMfJkqzrRtDjF/y0JBzUt0BqX4/HW38TxMGUmztPm9zv0N1+tZW\n",
              "X+A+NX5AaapRcW3hCRmXnzk8I8yvxR+oPKcEBJxmM9DdBi72bsKQ0Bp8+SC8DBV6MYOaj0uKgsxd\n",
              "N0g6Tb7VcTtlxTC383iKSDLqjAM05d+3ntEuph6Skguwa9i0t73lkMM4uIflMYkAAAMA12XsC8sy\n",
              "vVg3b+QhWRk0/PAecup9hJD1X/6n8Rb/XGNXArbrCV/8AHGVIH0UkcfJJoRXkXeNNQGO7kATSbZv\n",
              "sOgafhzax5Njn47pkfRyLFgtUUZ7XCwretm3WmTIvw/nex1SdpPFNWtFWM2ncjh9bxPJY86+DRtN\n",
              "8Hmoj336Vxl8+GQJehHsMI2LiHZmsuYDeJPSPckQ4WxJBb7Ja22swE0YD8mqr5Gh6Ln19p29IyTk\n",
              "w33BiqJ5V2x9IXErC5Xe7xBsYMmsA/a3ksCHLHXtmhs8/SQJR3k/ff70++VezD5kcktHrGpDCzAf\n",
              "X3yqqqPj4A9zG6YvGEojYEx5t7Yd35XuCb1lhPt4OPOAIMzF22lPDS+L6JmsrxDy/w/0jt2RoDTy\n",
              "tYu1jkRiSbJePa/GOVJwOStrZV9XkekDHnoMybwEWbdg9yVs8hZMvHX6SN+kntfjHKxa4wXymwvL\n",
              "VbBoLeouOxqVOWSg47gWSFMG7tpxgbRrX1xHWS2ccUATKcqu6xXEp7aA807L5gj9hDcQAET/2fRj\n",
              "C/p3mwXi+FuqMBLripEDNabEDDenHP4gy3OLsdEPXgAY5/nGnzcyH3uxZCU3ZKtaFnZa1qs0l3+7\n",
              "kpvHjO1oKRxR/xC8G4hKH8U8muP/9H9OGsHyxbAgCo0JtmGQWsvuKaD7KsBH7s7/3770T01q4Z01\n",
              "zWz9sVhRcW7hJdaFch5qWOjnuv75QBNZsbyye1949av7WNzFTQddUBTf9HjkRJr6hgYGKV81YPnD\n",
              "OtatHxX477K6ptjDTkdRxec8ZJvoiwDhnhygwmBmGF3AQHoJhV8xfK/EuBoOInQ8aCfPViBr4lV8\n",
              "/ZS1pIPgZ1fchhfSM/epkfN7tqZly8fkYL+5xlAkow/gGdS0XSkrhKF/37H3WHgASqqsvHx0cfDV\n",
              "75NEsJvrqKofKAUPP6hRZzxDNu6S99n766T2ObE/tOMOILnmtp81OGDvPJ55iN/5VkYRA4uIoYUj\n",
              "N5a+4HmWqXxHZlUQcfmetECKGIoPevdF8/aqrBwJXFNPZ3yeHMmL9NVMRk/cpr6vOB9ciqaOxOQy\n",
              "c+XlUdo8NRwY3cWFmIu9XzB0QlSF2za6fQC5bfNdXfeJf2WWerlX69vjGhqOAQ3TUYr2fBk3BU0D\n",
              "FydbkxLPpVD/vvEgf3evbS6sPOb5G6DwHPZyqdpzjz/+f52U9vU03yBFklAi4X17faMd4DzVgMkO\n",
              "7nqFeR2d7/B2955my+wq1QfMkkJJq+k3Zdky6Cu461b9dwSSK5BvVs8saZrXxV2RbvvAAk3JBX2L\n",
              "LwYA0dHrLJXBXM3U69IchsfyD2S2lRfV4X+y6LTtdzp8C/9DjDtNKFNn7YaxZH3oqFznHbaYN6z0\n",
              "+SWG+Bp6B4PiIY8KmuldGo+PnhkD0mhbfVaDaDv2rlhcsVxhJCJ2i8VeQ1uFWjN2aBPihhriYKYf\n",
              "6vINHdGICGgwuz6Ts0jx9006wRcVIYf6VZ6tPfoifIZqNFL57rPWBySjRHTz8s6nqUgmt6QA+Ke3\n",
              "2xhC0OXW2cSuAR+OgAJlPVfgABCxAAAEl0GaIWxBL/61KoAijRhlAKJDkg88GDixyouTaN6qewPI\n",
              "FvO+/gyMA2Tt9Lnv+u4Kmlf/j6HQ7pTA4eMM2i2mSpzpGK9B0teJjuDK28Qnn91rFY9cR9rqMsyn\n",
              "R4iesb178rKEzSrFi74QPrYw2fS6KhqS24iRD3MPewwiyB7bJpOXwl3CFae8gnb9pd+xrczTDfQG\n",
              "U6uibIde9VQGpj2WDFlnUkIuntiNCS5wkGNL4j3M+cU7nU6RWPxHziFv5gf7u1Y+m24XUn9vhmUM\n",
              "losMYgZ72miOHQ9NjQvtRai31PgwGAAcPq0UYeSyXSHQX8bBh6epT19GrYrSXu9+wXGZ9xsdZP/j\n",
              "XiKSQzP5ff8eut2rhKvkncGxTmQ8XufWyW/562mtJn1iQWqf+4eGdBAfyQsFv1OWVVllUs0PKXE9\n",
              "ZZjPANcyi0+cajmO5RylVn9AUjfjmphmFcYKlxFCzzBy6aKa59mvNl7HPVcvlJoyx5VDozU374lq\n",
              "DwEluozs6aqQQMrZ15pY8qTg8uvnmhkXzHP/HYRTRMEhPotwqLaqdjUaRCMbBwchXLRUPPC9scRH\n",
              "W5yV2TFH+61Z3v/6eFoPDIce91XTh2jJmMW6L78KS3R16mZBXB7q6bBZwT+XJS5IL4xVjIUlBQH5\n",
              "9kFFL6TJ/L7yqj/A0ioDquxaZGtr39YFYiKjfqpqou67lPh9AAFqWlcY3buNtUWYSBPqNo6HzC+l\n",
              "UWNc9i/h/LM4VOqCcfDtS+HKOwZZ1XAN1AyL3Zrb/OEVX3hsuQwIc0pfkd1sCCPuhmrevw11p52H\n",
              "fljHizLagiMkAYL1L3XV1qbY9uEogCfA4D7mCrQclkWhhiYmaL0kBnlFAfSjTw26J47ns0phDAF7\n",
              "TKNhJfNTQUlazA4OzAQU/Iqve7m+TKwv5rjv3yfY7CV4JEe1Z5RTs0BgemQhO3XynGBjoRAiXUWn\n",
              "iwk8gpvYhNF/Dr+JWpO/NvfOA9i8zba+q0rEBkwif6H4QBjStXm7ewCDqbMu47UMs5+9VjSfzAco\n",
              "S9Kuk3mfzdgzGCaXPTt4eI781rFqdXhxlKXyuJQA8CFL9HAmP3Ou9g2NooTjdPkqD4uyXaWFi+Vn\n",
              "89sGQyTWvACTgJBRdxW7H7PQlmx2X90A3gAkD6P74Cvsa7deyvw4E4rYInlb+AppJd3ZZdeh81i8\n",
              "6rExjN+XoogQX/F8PxKeCpP/k2kbAslJ2O/2rLLY8PGcgmC9DpavT0zoulIfspaRZ1FvPgcYzAqi\n",
              "VrNfCz3PxX2ljheQfTeHOPzPr9TrdW1JYBsd9kjKFp7l2k16SXGXbMSQsT+xXhWkn9wi7V3VKBXX\n",
              "Hsk8NXmMAdQZ/xAvOQAbORHhAD4k7kQvfUjw8ECmw6zwtbworxPxdJUxFSixr28lmPV5oPYXzeRQ\n",
              "Ju9oYBGwjsFha6OG0zrwsvGB6I/TBhLnXDUSpfSH+mz+GQNCGLx2O0QnINP7WdjroWRjNHcIwF3y\n",
              "gWX/OXEEaZPmxDf0Wt38WYnA2/PrvCSehu9DjciALmPDNqT9mPOAEjZ0t1pIrYzKbv6aEnGiMfYO\n",
              "uJKAAAADzEGaQjwhkymEEv/+tSqAAG8w7AVKAIQWUlAWT6SYpii8hkc81HED3fq4hYgVoZa18sUX\n",
              "6273bvKQqtG9yn7Luax1ibla9w0nJECnHu74Oxwkj5MlCSCGTweqlkIIOCmP5gPLMmeDdK2zesEf\n",
              "FsAYoYsCwfJWk0LxXi+NeKpi/u4BDqSVcbs9vaxHIiadRjDD98oOecP91TPupN1vvaQz1GZmcUje\n",
              "CtrJ2sKG8ghloBTDPP8/ezKz2F4w+b7JiLfYVyrn0JknlAizD4jZuc4+3Wn7X21STd46VoeS+XXG\n",
              "OCaGzgo6owE/PKesx7AwMsDpuI+uHGoSq6xT1DDd+dQ1K/JK1ObFz9mzCHbkcz5TDncEJdbCDwqh\n",
              "fDRqWwviCvuxDDtJuh1sc/6/2v8KL2h7v4XIgKoVSQP9ca4u3UgOMMMfshNfFYEuQkBtLh/FsSTb\n",
              "21xGxXf1Wf7PZ1/2nZzhaDEj5xDtylWv/n8sUnrJw7Zv7nV3k/YcfjGV5z1ONbEQej2DEryZ/xWP\n",
              "EG2WX4IHt0dkVnsvpiK9SIR3fX0ah6Xb+3M7Aq3g4DdfD/JeBFKsZAuSns/6tLJR3A4Mis1DFpCL\n",
              "ZpD6znyOz9U1ExlzGp3yQcRlpjk7ZNS50Gekdw/wkfnpFHiDN0ndA4fXQdDMlIRVg7SPQk0r3uy7\n",
              "+wP8//iUmUwM6cbMwW55nAnSwL7PCNGKQaJHO4VAIZtaUZMdYHuRDeyx90oqeS47QjWnrSmlLinR\n",
              "xcU+Q8eUjH7Typ9Bon+4keZ/ujopUXdeazbyePqxdIw4hSRwB34IMRCvdpOjyB6pUMmJS/tzmtAM\n",
              "eMYjlq38rjq9Xue4q3LqNR2Z6i2yDG/ng8X2vzKXD/at0fW9HAQGfeSLCP0PWFMIK9UkXu66Zy8C\n",
              "auZnxIz1GT5PNYum1d0l0hgcOn0tVgBd2lTpEcoJahDRyI1IHheH1sYIdUR42JNYujy7VpVE08Xp\n",
              "Wv/0jRCDt/cjF3P7q5c2L5k+fglhOucj/srYG8yCwOC3dhORQHctTc8EZMiTiGyjJSYGuTV3S8bC\n",
              "rlBjdIXeEhs6sUSH0gQpmJuO9SykF9RbBxD1CszkhLdsGlLVVdBjLZWDIn352166+ll3nyJxseWA\n",
              "eZVHzOWq1pDmElWcLa7Bcyah90XWrITTPVyn93y9mpjn7cw/VwKyZZXT1ta4pEMCRJKOhrW0Vzqb\n",
              "tAXXbZpvBzSIJdf+ZqsOeJojhtzyuTIeMPe/24Biccu4kywV6yEzCIqc85f7PiA+YBRc4wr5zcBH\n",
              "6cPQiBI5aDj5pQAAAydBmmNJ4Q8mUwIJ//61KoAAbybttjYt9lMALD5mgRYqTvratGPyeGfBjTaj\n",
              "/eQ3UHdKH1kFfUbAybnsjXTvyvmeRMyv4ac5r/hsm7/l1x8+K4aVcJn4KV/JTAy9yEouw/eh4vox\n",
              "LTOnQdvDUzbnplXVYGBg3LnFn9paiAF2Pp7JGEcyRkaT6RlOgWV3mP38VjzMrM2ZhOJnxKtFjiNt\n",
              "XOIEsW+tLcGcOrIXSkbKC7IXqpDTPn1ysBFPw8Ej8fs1rnR/GM+7YiPA0eld3tY3wZFTcHSXe7/c\n",
              "MIi9LwcVVBOgmk9I40OexYsce1ixPgruv6uu/hdBWgwEuf/0ZzGcRvXO7fiKfc2t5wpUiEzFgO3V\n",
              "5/n66/PS0EHL4HjM9dz87N0AarGx/7QqXzP/sgGMLmFqGsDHOtwtib8Q0TdVeo36OwsKi1XS2Mfn\n",
              "9/yYtQDvv11EQK2VcJlh6BnBA8dKbOeFJDxDt+B04nYlz1CM0N4YwcM/YUsoubWkIW5EEmnbU9+V\n",
              "OoI/VJXsGubSIpbvKdXKg4ONby2Fst7vgUu/9Dtag9iOhywk8k3oyXp/c/nPSWCEvRSyRld1JloK\n",
              "Tx1Q2Uhkh4HRfMm9B8fBs9xLP575hy6zI8Jr9krBYYZS6b8q1i7ne2vRMtKgkUjo4konnXDE5FEY\n",
              "n+y1hxvrYK0gSO3Xj9VuITEdbpx1Vf3Pg9M2HgjNMkhfc5ihGOHWTppHWeTwdqO+kdO7JAQgN7rr\n",
              "DvdfxxjvoFQp9R5/ybhv3Hm9y2Kd9uPUI5xB7u5VivDQ7IxlkGTKV6BIGIZsIn25FyJ2qrqyefzQ\n",
              "AVaHFUGro028/gjsZcl9xNt37eGH+2ExhejIKF9+SB7KMZVX3BQ9xFoM03wlOzBlb4vLqG/KVDe6\n",
              "+HOdKNVtVWxpPxyOvWlPahK0HXrrTxIp+MoOw9enRP8424iXyZuThgA+GiQ3kiWiX/60z+xj5Rrz\n",
              "KBuVkUeBCI4BVjJjB4v9u4GjBMgyiVAWuktusRwD0XTYINHTWlX2q5+joa8XgJ1KhK3CNMs/xeZP\n",
              "vVDyRgd0r3qDCPRUlMar4ke69HOiPSAAAATlQZqGSeEPJlMCCf/+tSqAAN75SAHigBwI5UtjZ9Zg\n",
              "BinYc9OABLGtvbjqPLXlKx+dvpqsrcz+v3eTthL8IM50aOxX8XlPaHx7hye3m6AMS1Py2C3zJv8O\n",
              "S7XnDy5CfPWIBkq0p4crbKoTQGrTkkdPOGsIINAeiB8t+QdTRBV1UeUB/1LNuIbIalPpzAArBPqv\n",
              "0pTVYlJ94w+tA5OkfKX/dEa/efCPpX3q1iI1Zm11n8LqkBakSj/U0m9A5PidlmCqy1NGSwqpxKRz\n",
              "WGia+AgsHtlSwoL9Wef18sdDQBRTfCtCE0kmLjYMgqSwaEqBHYAWy5g2D/E5FhO3Iw0l/kyosMV9\n",
              "qoLSozOGji98HzokDYZXCdxyN6YsIU/sjjp5x6NA/q8VttYZoPdnjct5eWD5N1lT3/EXgMP3pbyD\n",
              "k5A1h4GYTBCs7Wa6BTwgXUbd3NxYrEZ+Z34Mou5IAvnirPjz9/ZG8gWdOx83bnS0PQveoMpUq77M\n",
              "NESIwhqALOXyuBIq5QVnxKd2Bcw6td3yRh6bOn8Auj1ytC2VfmtZn1mydtibazF5Rk9VoExZiPOm\n",
              "cPGqCa7aclneUi9+o0ymkEHn07e+4keePiFz3DxEsuncxRHB5Cpt/XKxI2RGhDWfCtRJUESV2ipC\n",
              "pBZfpnRG9ZUQBc2ScGli8Q3ONAczJ1B81P+49NZB78YoNUhzqgP+uNHWfhsOOyFrxD29npEmGaZH\n",
              "GcSQXRNTRjUnO1YgkBt40rvDn2hfBONl75ZYUsZndQKQuxfrajCufXCMWDOOFDXS7Cjj0hnC7tvx\n",
              "7AAZ2weGie2ssisnoCngqOG9/JtPRP6hcLQat/z/keHhtagAnRdxsEbyaFGlUsiT+PY9AFM9MTt7\n",
              "fTIyjnzFWrDmd8RYyttZtBLtqFeOrZvnLMfF1aBIW4agfW4tK93J4Jny2N6lv44yCoJOS4KcRzLt\n",
              "4htL7JSzBag5iV/xQTvGt0pOF8gNEQ2LqczI75xuBNnJ+P97zE0XCg9YTnxcQ3P2MMuyOI8htjFJ\n",
              "UeKg6SgcsKaPVsaBYsfGWKDb+hijYoA4mdZ8zsKuNxVp1CejkGxSGueR/5JZNPU0zv5M06gXSCAU\n",
              "7TSj36kPWR8N+GUdAl6uBizBXU9EJ58KOl6mg3jEGaj2gmia/XV3TjrC23VZotM+oACDuU6/D/uw\n",
              "GnpUN5rvlCg43cyj0bN1dVjOUJ7R5ubdsTAS7DQBBB9TpEfep5pJaOMBnRbxwqVNzyhT3UqnL9aD\n",
              "DB2eNOqsFfDsRrfvHkzEIEWWVA4TJvMRPyg6lWzkkZtXpqVqTT9zGq3X2Z47cTz6whiWvUoZGebS\n",
              "RXryI+vTOln4I9k4etlbSpzaHo9VE5yQGbNxhpgeQaxM27GUUzz3aDriCSRpg/f0dCr+f76wq6Cu\n",
              "fnSW4PzeZAWxdLef1C6YX4Jf06WY+yl0cjgteN4z1+ve2SicvB5TJBIMAAuEGR98IDR1RN0iL7oV\n",
              "aMS8iSY4DkM8HiLxlyxdBxwz1V+Z6KxD6EWYIHQQYM/JqziCr+VrGU4aUAnrOAvREjrXJB2yndiT\n",
              "CR9GxxWmCnkmMfrU3OEIibOMMIm7Y0Tnw6eUoOwquuAQytzrQGnzV+uWrWKIKnDsRqSK7aUwgrMH\n",
              "+7KPUruxcT6yJsSmMDbGOEPpyMMrbR+KZBEAAACvQZ6kRRE8EP8AAlTjUE/ttPVhFXN5QS7ySSVu\n",
              "WgBKk8OHnVRwAsE7UpkEAcli1QDS5M1YcB/VifKP5cr3pQeZBK1OvxsOGCJa8Sfsy7MiTaCFJcHQ\n",
              "wcuVObImI5o4VyBI9iWg1A3kgk2AxmMi01YKok5dUbUxsPyqsYg9OTSSuw5Mf75mnUxrCd0+vIpf\n",
              "QWq7rVjAqh0imPakKn8Kf35f/lWnYKR/QJQAfJqWYJicGQAAAGsBnsVqQ/8ABUEfikE/E+7DNCx5\n",
              "grv285QqDaAD9fb4XEBe69C+8LONQuNqCBiQD4fg5a1PtelaiLp3T6vccvJBxVww1Dcf6VCAJqWO\n",
              "f50brysic7uttH/Ap/Wxv1usvhZ39grH9F86i8w9oQAAAsZBmshJqEFomUwU8E///rUqgADfWxBe\n",
              "F40AL8grkN0DIi1Su8M2v5dkcZc/xMYLiMROSJjHuCE67di3e9BPjoLqEprpvbf4EHZE/XsIPKzZ\n",
              "GIOJKKDmTI0v8gOZ1oOg0H2Q0ZdOPDcWUkGCkPVV4SP0rWYw0xMUoa77YjdfD4i3PUxBzLeAPRk1\n",
              "FtzmjVoQ9BGq606BQBepKxn44RT1w5u/zH5Uila+1b22xReYTwb/krZihszPfpVa2LW5KQbZvMuM\n",
              "yROyDcfG+X7mmXN8iWQfQmz58cgyCzyhZItzg7n+oLzbe4uRYBsnoYByyEOsPQEsZxiydvAK7KQk\n",
              "4p4ZlCvdgnIMsIKjo2+hUXxEKN3XUOh6IUJs2mSkUg1rFKMDmXeX54J7DT8Tvq83F72Ma+8qgCG7\n",
              "4UFrtMBH/pQrZ4jkHe6Y85b648L60GdxlhjCBKCky2mJoBMiJ/XyYrAzdbnDID7t+2Gc/g2360JD\n",
              "L5s1SpK7l/YUTCz3JEAhBncDctxdeRF+jGJ05jjiqbkLQOSq830qRPLh8KYNSsY4iv4OtEE0IS3r\n",
              "Fq8BQdO0C3xltZ0NDC4PiR22qb8+xpYCFE8kJIfcSphE9hjCmVW1sWOmae09KuWnk32lMg6KcvcY\n",
              "EfVZ4K21GdW2v5vfDKrB4syBgBObjfOBZoLyZkx9hwJoYZXbgmz+nw0qPCaZ3zSh3vb1vZZbi9w0\n",
              "y5GKT4/mojODJaMO0n+z6oH/v1Xt/WO8aRzwPzLXXJDmJEovQuPHPFuUOoU6Zpx00MJyAEENhE/s\n",
              "mopONCXE7Np19kZ0apEh8lZAqKws7iuI+31n7DO+d2LXOjF9YzTKVbOCdnSsIWzMWwZ/twB6QdRb\n",
              "6y9ujH/Rhb0sWVAV7O+JjyG5CDluZNQ4+RS0uFhhyMfVuPbl2tI1G6DeCcLUpLtkP6V75SKEQw88\n",
              "IkTK4QAAAGUBnudqQ/8ABUF00ouPmxO0U7D8qv+7TLk0GIAWqYBbiOfjhJArbpnO/CxM51OtSCsP\n",
              "1tEhsqdjaY5+5vdQ+YPPmKdxif+RYVyuJnrd3wVeYzJsL/zZmjUlONGb6k81i+KcRkhiEAAAAuxB\n",
              "mupJ4QpSZTBSwT/+tSqAAN9bEF2NNACKKTDO7CNvx0TxglE+Nk1JyjSyPn6GD31qqgB//SLTsAD+\n",
              "Z1aH8Uq7uXC474TBgmbgXbekaN4iIYw7rMluvqnZ4H/uxy8IvMp4KfTaMM8kW602ng6byKtVYtLt\n",
              "5DNHIz0barEa3DVNKFbF08fkpWnqNZmfAPu92U1Kuoz9LW2ouug4SHBJfco/R37tLL1ewxmVzCeK\n",
              "HsTBzT6b9MuszzqVzeG/DY/FT0ddANMYUjJaQBpll7ihB+wSnDcO7sAGziWzhMK9TGw1Kj9Jhrwz\n",
              "x3v36P9Z9AchCEGNqJk2f1FoXdynPDRflzB3n86/HeTF46ij1TqR9JgPR7v/SIvrk+rjHwcmuMJ6\n",
              "Og9uLLWSGr5jmSB2dgTr3Pnp2lFqJTWOh8ARplbPw25MKxaeOSAkrbTbuQAnS5o4FJeaGsbrTRBI\n",
              "HXAo+5pq8Ya+B9QjxnUDRwGNDL1cqHt1tziRc/qmbo9fKcf1/lYF8ObUUsuYhASQpDEbaSm8A7/E\n",
              "Ln1fFWT0GXxudqgOoL1ZmueV7R4Sh2rlOPMzLx9ZEaTCKNHvecQ1vtYERGoudlGebS6+zPdSg1we\n",
              "oLNB7XJoeW3a+VIHdGNKenyfPps//zG76imTckmsMEEJlSBnC5JjizkqkkNl0/qP8j6Si2CgZVB7\n",
              "v0IcIcFmaP5CDMWZmwtr//LFLVm5QVTzsqgNhJcVteULmCUS0UExBP+wBEUIwayFPNO7VQ0UNaWA\n",
              "yZkYeMUv46qlnGENoWAQKHQc37VSfac7Qf7G1m/BscsQOi3kma1WT+BHryEM3bIqZ2UWQoOm0p/v\n",
              "5qlyBT94hSxBMDrdzM/L7OgdUai7fq323bc1xFb5SAXxuLmG4bvCt7rMBMdRp1v9hmFgGloK8DbX\n",
              "Wj7UCkvPlWqA9jpeLuSIn4ktmZrK+tcJCjYW9cDyNhC93rfgAX5wsK5refZRcCSi/nQoYJdWxWhh\n",
              "yrxdjYdgAAAAVAGfCWpD/wAFMtZxkLymrkpOFPXbPfCrThU14+kZ6Pm1Wzt9Olr8wefIh6BskpG+\n",
              "ZuXX681p1Nu4IPLavQWffMV/z99B5XrGIdjKcLTRHYQzG6TrXQAAA7tBmw1J4Q6JlMCCf/61KoAB\n",
              "vOZ4rEgAiIA0aZ4fWIwpe/gnvnQbBZntYrrLVvCDrhmSq/iCBtBX30hc54BaKjhmeoTx6BCymOUg\n",
              "QblMdHjJImWguTbEghkYEyT1u4eGur0OP4Zr9pKiqeesyW2lSfbIScblHtiNmr/IiQArlrvXhNba\n",
              "BdXiH3YMyoZwTglMCK1GUG24X5sleEFu3RyVDew7sUG0CgOlZWH5VXZHjBQziUQniHOc10jUotlO\n",
              "l+/FSPKRc8L9nEnBko4F35aS2nXd0fS2eF+ZM9kDGHibZgRYPVhcrTg3KdpiDo35oec9hZ6WAgDY\n",
              "p2QVzxcks0Pv5KRVuQE+FysLIEGi3KexerLAYE8Baj1o3SXtNoAE1kQgrRp67a6Guv8eAs9rIJZa\n",
              "lIPBpixbnWbfUu30Y/em16QBGOWn0dADIRPSY4U9v9lL8d/X3oGRKIHiYmMUX5mKAoG+SAPi8xeK\n",
              "1rS9e3lkMzUfd4HOTAufRyZdZUBOxn6CDHWjiOHxPRgN86aPedaCTo2oj1gPOH+Rgv4cXW7ANZJb\n",
              "GzBY3H3WvtGqlybB2l1abAeQvvBM2jboITSEicz4hzTpBwXpEiisfCv9dq+V+VgdQv+PdnHR6LCv\n",
              "DX9IZZT4nd9yCzdi8d3bKv7N9rf4VQC+mTI/8+OonbWAQMC7z9HXjheh3d1aG4pfjplEjm+FyGUV\n",
              "jSX04ssVqVFsrw0wQOh8I8jrYtz9gIGDNKwdVhR136EpxO6+L7yTxyfbH31/Nb/rpqpNwdvREHJI\n",
              "A//9oS02Z/COOMMsJXWn6X+h5jfGfT2bi4TxpSpV6aHdQrzG//Kqc/1Ajr+G9Zrw28TqiqEdvXo+\n",
              "UcoxrUEbdWtiS8dBPRS/XoxL0zf/ukpzvji3DgRNbtkRvVxRmPBzA84OrKXk80D/v8mSj1h9uUhw\n",
              "X3WCtTmNuCnITHO8twkGfl5odf/zJRu1Ikqbg92LhDsOOb7diUPRgCazAKF4YU+mb+rqTELJixka\n",
              "4YzAkB6krauG3uEdZCJ7wH7FKXyZQvZyttrXwtqoZStc1/BnwUroKzYRrfBNH9loFcvPpxU5rQEU\n",
              "7p9Ixib0eHeA3gegfCrR2Yo9GwLoVdaIouqBX88AJXpTQZl83hOiVYMdTt8YxgAWAm/lzjdiA7E9\n",
              "ATnUzbRSMijyAB8Gu5qVtug0aEPkagzpo4KaKwi9prAsRgJCwMobslYZZyVARD0N1G//aqKXjNKg\n",
              "8qaJBvVOhtVNroiCzRiA2vv0ullVCSbUUGJFAAAAekGfK0UVPBD/AAJU41a8J2E+uBmZ5P9I3QAm\n",
              "qR4H8hBhnGa8NIqaDDXW44g7bpgu/CbnnhK5do84kKLcifD+OfSRWHriQMP7qNKS1s0W3c25GX0c\n",
              "s9pN/BfgfM0f4GQvnPugo6syzJ6860VWtCwWOSTY9Vq7obgyHgnIAAAAWQGfTGpD/wAFQR/U3udC\n",
              "6r+RfthHqBmk4n+na7R5L6BOYASVjz8lICnGX69wVLB2pPkQWjSXZstDSfrA9YADTedBwSJvUfIL\n",
              "lOGqe0It9HuxpfXZHUx3vFLpAAADmEGbUEmoQWiZTAgn//61KoABvrYkvFVlgC/cOcmLB6XbjQs1\n",
              "exVNSUT5XNpJj2cw93s+qGHsFaNJynuP4sC5FkEERjHC3enhy3EKVElxuEl7hRU+J3CTBTK7Mapz\n",
              "Q8oRJGDSBtgVrrEdODYxtPZbCPRoAqKYB6ZjTpjy8VSrMjyolqFoITgzQdLhrYoVoewJmcU1OSJZ\n",
              "KK9lhWGlVoPwQtVObttFnnFdFziYCkWEwhROeTFz0CscD4M86EnbaiXGeZ3UuAMzoYd5oOlegNOg\n",
              "bCoyu9d5nvYTX/gVF7p3N2OO0CgIKiMTdUllEu1EoZ/EkhmkYI0zRf1tkQMVWrGbGQz+Wp1HRN2K\n",
              "QNljg72aG0v1TaBI+Azv17HYsgclXj3wqZUpffmpH9gLUAXVGnW/WqRouDu4z9WhtisPh5ziI3Vx\n",
              "ZlbzDRE7pNl0/1ci08mAAjrRRwcpLyv6PN+RuZvr0pLQ5YYBsJzubBoEJr2qcHMRc11TOCqcgbFN\n",
              "KasNQIH2aos9fe/TAqfoyEiQvrgbhYhN8sxPxBON3GZKysrbxv4lEEe54BPIbxN84q54VkQFIR7s\n",
              "OZZ0ZrgaY35UZ0rXeaOgMQu9Lkn/HSyy43WSRQ7FLyiQvbmQudyagHvfOPKr5On+U/mk3Z/hbk5x\n",
              "lTyigbGUC0fxa1Z4XaX63JnP40mFEHHeekK1fYGGdFsYiPNJHowHiCE6u4zLu8VMeXIJ12X7AhfQ\n",
              "c/tfPlrfumUn2XMQToRhiGD8MM9ilFRsm/H3E3ETUnxNwn0NdsFkxygGbXqO8I5YjQWdm2UFkohg\n",
              "8Vq1u/J5F/D+3J1Zg/SvtPguy2LTf6u0l8JqfQHdD7gj2G5NNmoXJBS/E4Z14wmtxAjf9TgqKkw9\n",
              "AAysENqRAvw3w4ABXzNRO/2xZGDtQmmAdllmrz4WfJtJB8Opz5IQTwA/64eOnIkAovDvfCx1jkOF\n",
              "FrIF6eSQfVwege+xd5ZoUZcWFCQ9sRPJDFVyJlav2ppYaL3tfM7MfdvQgqrc4Tv8oZqolpYqgl53\n",
              "/42VDTKWt162T3AnJ8+wn+EvO79B//kOVCkacl6X9VjgakLJB037gta+ZOwjZCyhyccMU3DwctIH\n",
              "U/jpKL/+6j6mIBXGo6cBWpVXlXLCTt9+9WyZqA6kM17A9yiEFSYv0BA0QYVXaA7KGmizgx1GcIl4\n",
              "xD6eJ5nWs7T88Bbr/E0Qv74q1w4XJjKedcGGsC2BAAAAZEGfbkURLBD/AAJa2y0V9whfrkidwNdV\n",
              "7XVMnIJRChbIoCsRICn9QHdVTmAzjrVg6+GOQALlnAIyA8Slk5O6lgzeMrnwfG0hh+8+f+Vooox4\n",
              "vBVl5mcASjokxhllLK+JW7APJzUAAABNAZ+PakP/AAVBH9TglHOjTWk+wUWR5dj3K73TbXRQqBd/\n",
              "X00A330bCy7SdeWQCt7/ukWhTSG5H1g10fVv0mjmIxdirztMx+VTal4XJP4AAAJdQZuSSahBbJlM\n",
              "FEwU//7WjLAAcdDtq++kLACk65ex452ok8o/1DFfd/h6KO0fI1FtsifIZdsUJgAcYZs1CzcAZhxe\n",
              "ux3TFuTCmpWiUQFs5yiAwsE61YsgyzWle9XSOmSzGxm0WQQpTSzIldwY3JoZX8yyg4VMgYSrSTaJ\n",
              "0tZv4O4ySTmjorqH80F5hFa74LFbwGuOtZDgTb5qgxj8o418SZlvoDcs/WOfvW/eY8fXx/q/ycSX\n",
              "HXvPZOyNL6hs4WvW+F2tzcUcVLmrn9vAnRX8gP+MIOsdOjhydXKYKYLHKKf0QbEEIWk203y4NKp4\n",
              "POUnv/kT8BTj/9X8QVE9zYvHsZiQL5VozwPIFVodtjum8Ik19tZjpwya0b1UGqPZws+x0uwQXP9P\n",
              "oevcWOs8Xdl7AjYnjlL0fJ8gz1R+nmabMR4rsSt1d2zU8Gj5tZkaqimNfeegXSnvEe4ngttX1Dbt\n",
              "jJWcC0PnJleb24jhsJOsKOc4PB51CewH6V195/QeU6cVY9RxmErMsxL/YVuojpKa09Qsvc099Q1B\n",
              "+78EKGKrYBTg68Tqxl8rq7tED777lDbaLoGMziQAs9agWvyybwoiXqc9+Tqte1KXtiYpfOBWIrYA\n",
              "FjaTbXEIfY2uc6bwCYDo8/oZM+ZyBhc0sZqo68LCRsFTEkixHFN+GjoHJTatCOmZPqfkn6YB2fTf\n",
              "3CaTGOjVlY/K78VlxEk1Ro+7LIMNCzPP04EUPrm6Sw7pHK9jH6H9aCMD+klHviJbOrYZbvEfIbOe\n",
              "n/3VAcBknQFw1OU2w8B19rCW5ZtEsDvn9CgAAAA/AZ+xakP/AAVBdOlU5fKa7WhzTlNsPf4VN+wV\n",
              "ejdoCoLys7QzkbDxDzvfQ208bZXLZTPbbl0FSiqJDtbFBtfBAAADV0GbtknhClJlMCCn//7WjLAA\n",
              "46KQVDJnAAn6Y85zbZMuSO2rlE9t9qsnsw5KJjAh9no/okxfehlWXTW5YHjZ/a7u94ahlZ9Dx2ej\n",
              "th00xR9Qv3vygfYEkzPdMgfQ513+e75YOxfziN0QSctLSSY1O19/gtSXloxSqFYlG0EyLOkYqoOO\n",
              "udQLBnlG4Dt8lYelz4eAur/UVDDqix6I10RPXB5HYkdCPLVXih8IpGCOsZQB0gKCv/KT6BaeRwWa\n",
              "wxyMy5VoaSc5dYY4VFNpPSsrrttTXJd63l6xrZYqEg1nlE0KNMI/E8qAvUALBSOBn9Ld3p3NIZmf\n",
              "kvQEoqOr0I5XIjn5yB60+J+N932CFz9kIPIVtpSKVWaWJTUTU3rMddZX/PqXLDcADCgzoZk7wpWH\n",
              "EKEy0XYJ2fesg+svPiEQmzccnHSjPHCe/PlWxIDyO68Sf62Sxm+O1SvnvKFg8dDEC7q95O+vFPa4\n",
              "/weo3MCNYC1ShkUohwxiVf6foXO5JpU8UXOBOOD1Aw3SGj/ijTpf/3hFbNCvTsA7U866MJQEdAJ1\n",
              "qaB18ZoBxfXvQtP3KxK1jej6pYoUKuUNxcgd+ZjRo3qcvjuUpBVWXTK2PoMlGRg9ITcmvAPrbxYL\n",
              "op32T5ryheOuA/DF3QG/clLTLfa+WKx/tOv5fRvoDqXo8RaQyzKXFS8JjnZfvBYnDsOm4FxwAf2g\n",
              "fneviTyyqFrz6ZZoliqeAvimML+myJnbuIyPPYKrKX7HUQUM13zQr+cWjLJfCusnsUWYJwjz6kID\n",
              "3JfOPo2WXscdAQkpQNXlPRYjWTdfWdBs5m3/9OKScl42m0oe6PuI/JdxuruuanrmaBsvSwm/+voA\n",
              "KF6YGujJoCbA8YZ50R8ltqYVBNJVVz0Hcbctxr+rx5/B2vtu5ps/YWRrnCJlxznjY+fmB0qye1l/\n",
              "paDIK7LFsc00iAWTUO+QTWP/Mcv1rXTWnemf/1Ev+4a6m7M6Ovqh/CR34EKR8tFkCS1JzgLmNFuU\n",
              "Vi8hDkWeDWlWKZFGTAe54kPAvZTtxeFiNlv8NRJ1gqrGh9XHf15c5PVz30vUCgYAWqfCvBa4O0fc\n",
              "Ehkez/JUM0ZlAic6ouwwrWe5xfce3j3GZl8hQpPyZMGi5TlV0nXxUAAAAGxBn9RFNEwQ/wAESOQw\n",
              "PgF1r316As6aXRiGIPyCXJc7D/xZwhGVrq1ugMKAEFrSjFTXsgpgQALpuSXGboe6NShYYO9sVeVK\n",
              "jJxXGh1zyjEUgywm8dDMbeAdfYrtHPu4j8C3xhI+KlNGIG8KCkAAAABAAZ/zdEP/AAmsZp0eLkpp\n",
              "cdsCyJRK7ygA4mhmXXfVHybAVceEGIRcPxAbCLALchjjPRAHDJVDQah1BfC1KWx8PQAAAD0Bn/Vq\n",
              "Q/8ACZKigXHEKuw8P7RnG/EaQ5jBQGl8aGLqqo92S4ZJFiP95wqyOlmITFWA5MYT5ATvVlPwuppI\n",
              "AAADS0Gb+kmoQWiZTAgp//7WjLAA48dBaSwA08z2axL9u6BQ/P9XfQ/IvPJJzF1bPWvHpRVFJoc1\n",
              "5fs23hWrjz8xBvtgoLCYa4Adi46QlbssxnpYIMSt2zNcofPCEq1IMh8CDFLi6eEAY2FDtf8KxZTx\n",
              "unaSEH++kdnFfEO+199GtJpaTMbo+DD8VDOhPw7+PGOIF5yDQEA/ufKj4OgAHJGgc4YB8FW2Z3ws\n",
              "aBwG/6gW2P5rj3E1XxuQ1JXvG+uJE3u5a07mMCHaVJC7YMg6Ga6SzaAEoZmMu6V74ICOV/JeAncY\n",
              "WT1EpbHM4P5Q6E9hJdhNsXjg6X7gVt/UnunGAg2+BjrYfjmNF0tYJwaK7RJM5Zkvb6sDmeOUTQ+5\n",
              "v9WbvwYXvVydwa6UhyRCvId6X1ooeXLCBMoUGM2D+hukJw9PBFMigIegzOhWZ6m7pqlddfr8akNF\n",
              "xsA9ORiv+8knFfrE5C8L2fMsBY3TWmTe/nivvf0otbTpItSQiToVLXRaAwHtV1lV/raLis+nUSoS\n",
              "IAiyKHFLd7q+gEKc0eNb/YdbPmmyMGfPGB6z0359dkNlz9c5zp7SYqLBVL3tk09iYhd5cuDRv6vc\n",
              "M16qYModaqnZI2bSZkTBCK6z4h+KL5EbTLU33nBxo3kkQuVgkeFd3LVSo5Y3F+dMpoppweZCgHUN\n",
              "D8uMGMTojm8BckEPBCVb02+RNBirjx7k+yCPd1nX/M7/oPbBHrEEd4XozlryfkkF0Rq3ijk9/87E\n",
              "mzkdWYUu2I6F9I9FBtTWGpi8lGz8FwyLetgnz7io69kiBjx7f2QI/Vskw3VNlcfUc41z5WtQ61mf\n",
              "MXpiDQhcYYUd+7R9kHskZngjmfSXW5P1baaacKkMi76+7ZIGuIJWd5+IrADqYYPBtDkdGnJ4/RQK\n",
              "OKVUaZvSZyAOZwWa1ttsxodS9spyV1SB1vJw1fvPF6fOxk+CYfdDJZImWUG34vB2v6vW2Aj2G9pn\n",
              "RKl5PYpl3O/diHbk1u568YMIUvIOTsVM6Pnn9OwktvNdR9UhzG+AYbg3rK9TFSU+TjVVxcOcnceG\n",
              "mqKzAF86y3+PnnBxWCReccY2SKROMluYm5f+WIAg8U6llKQFWeizOHoiTrXzBGDGvwAAAHtBnhhF\n",
              "ESwQ/wAEU5jO7yejI4GAKwPweSXTIK2euk43tI7Qy2N/kW0Mm1DMT/QOLnmUHN4NRxf7TY8lYyZs\n",
              "ndh80G/97oYw/GHi4MdiYHY+Y5b+q198roduIbYj+pf4T7LD5w86hPQG/xK5LH+Ch9p641B0zvEA\n",
              "w0in6EEAAABEAZ43dEP/AAmqGoyKLkdSO83cXn0VKHYDXbOWErs+3b/ozEpuRuUnoYPoqJuWxTPa\n",
              "iwaFRzmJZElz9Xqq/IFKrRnpPi4AAABMAZ45akP/AAmp223l7HmJFijhS2ItEuOyNNr+we9b7uoO\n",
              "ZkmUe4l5PTvpofj84OUJOg6sxLA5W16js7YKDfP1mHgt/6B7FQ+QBWM0bQAAArdBmj5JqEFsmUwI\n",
              "Kf/+1oywAOOh23p4wNLACN6iGfFzQt9Maj64Q470tMcg63zM/roCypXLYp/RPLdyM5Kxs1E0bxqB\n",
              "Afqo1OsMg/oc3Jp9ZzQqWrRVfSq/8zQ91JZcQAwAXTL0cqoRBKU9Bb59ZcOJ/lXH6wT/MEKiL3xF\n",
              "UyfWRR32t4iKwbE8Cw3YiDSmvETcXNOUQuKQI4WwMGZB2Gb0gEpd4IbZ69qBWvsooXbecncSnUY9\n",
              "zBU4RCc0dY5sgiVvnxa1NS0mVyK/IIAWRW0IjSf5V+OsDD29E7kKYc3PHNfl4gtlLSgsm+9qZH7j\n",
              "VVpgZtuJI3pxF8mJwt9CYLxaL4pq6iNfzUZ7r6rTMpmYASIYg1cutA+Ht+a4WFAlZn7YnG1y0JKT\n",
              "z4D0zARGW+hLs/u1iZn5Ia7MMNFBtXHkl5JFQM/5+d2ZTSr1iszpeLDCJugEiPS4m24JBjl3dWJf\n",
              "4usQA9Qb4hAjnDOZ/bwOANXEK6Kepg2f59YeHhqWimnx5RR1aRMG6nOW5Sw1H7qwO6hejQetoYFE\n",
              "w6KwCvlWe89viaFgoCtXPbx/IstruxgtYwLTXtDBFR/fi455faNdHEfTUFD/NKfCufqjOSrabE9R\n",
              "O6hRcF2TnFQM+U7TBf6WUgdsMrUFgQOHn0L8O9vqEPog55qut9IW7qNghE1/9G8dMVd80PRS3Szr\n",
              "GlWdZUhYT1UXESyNLriilE6pzFXm3PNkdTD2yWzgtNrLTG5Iegc4DT5CAj++ew8hEXvmWNXBF8Eu\n",
              "vKR2IxMGJiyjWXsCNsYezL26hxM2NPr/0lyvxkzsnYQzWKs1bivLz31IDLZ3ouBXwaTKlbNx6oIO\n",
              "NKTo24g7KzjTM4Mm8WBw58kSBc4CPaat8yNolUAFOo7o3C/enj0u87QCwgvpt1gKktJAc6UcX/Tq\n",
              "oAAAAHBBnlxFFSwQ/wAEUUPLGVxa+hmBwwRIkEVzJ8dCZLO0Yn2HrvHJtHqvf93aHh5KUq8gHX3A\n",
              "C0fL7XTXVKNO01ZrqXvpqjRKLryB/F5xVqvWFoXcmA1GfyZclztiMx+wRjyLOdLXSAYnhOxipns3\n",
              "ZVVBAAAAXQGee3RD/wAJrLW7vQxJzbJZ2BZN2Kk/CdezlhURFrjWw947zDaANaWAFbFEbnMrvGeB\n",
              "aWnD/g8oFkjCwyF0hVg+Z384+xPlph3JQktX+T2ry0islSfMtYc5p5xrUwAAADsBnn1qQ/8ACau+\n",
              "nlQDA2PKibCuWjkKucBCppk6jmCDWaphiLG5alKNLJgBhVgdvPAfnULbMNHrQFvBqQAAAxhBmmJJ\n",
              "qEFsmUwIKf/+1oywAcdbWl93ABSE/wAxNApzEjm70PgcpQdNUpusR8fs0kJBtWZ+tyWTjpS0dhQg\n",
              "xnyzENLM5bLWftZpT4bRmM+S45QsCRW9WN36ZVc+zhAlsjWyqlz38/lEH11cjodaoiVqzuz3IyUk\n",
              "CefWOCzJspeNTErvjW52C/QtwzcmDYIqHa98+7W82weimIh9/TdOWSW92VcuLUYMr/gKetCAK428\n",
              "CFSuIaLVusxamp5F77EMcnkSiSf7zSqX1VMwm/O+sRs718PE9nQ/uDUiVPnHa8UyiFYK2dvhYQe7\n",
              "oK12ryZIhVAd2dv1PUBdu6LFr+FKk/bqd0Kfh8f5UVZOlSj5kql36Pj5TmoHW+sLcRTiR13hnIn7\n",
              "kdtdMGge/GE56jm19s5St2/WtGIcyjXiAtTlFfkC6quVy0DhV3bKzTba59Y2dbJ6H3a41juf2uzQ\n",
              "TcIJFe5/IQ06y4rnylRRCja1Di5IubKuYxkJN4tD5gxsBJ+WCncpW6dOnbtj+kpeglHM55Q5Mu0v\n",
              "H5Tl5lAtWQb7XyRJLwK8TpZAX7+hwF93Vo2KGYFMWkispGuljLo6CfDRO7YzXzM+QYEmQx9xKtCZ\n",
              "GdWlJc3VArEnr4L4SWJeI5aclUlU/qM+O8onQDviaDzwvduwVfLDWqhJtFux+G0G3Yx4rgrRPZga\n",
              "/2qaKMOm89uZeHduKdRNnAl//T8JQ7sttUSzdbb7DekPBL+UwYLC8lYz7oG/RPpnfjAlpqwRCSSW\n",
              "j1Bav/9PhI6vOrg88JgBXGYT/YA/WJRywitGWGH/CQbox+2Rb4X8O6L3jk++QS2VIX//Lb1YFbNf\n",
              "h+MN0ZTFmxcyIpIEmQVuHAFUxNUM3MgzdeGDyGG3RVSxOzBBn72pcoEaScMSuF7yU3Ny489fXiXR\n",
              "Kez++Q3C/K5i5UkZdt9uZg4NUn71hcckMCOwF7KNarX3yyXPaDyS9k3enPijR3qCgQtt892gH+Xe\n",
              "eN8cY1aYOOIdiyN9boYhcwCTJrOWnhxY5sbq/2z5DwVY7z1wNLXusq+EF5V4mwoAAAB8QZ6ARRUs\n",
              "EP8ACKp9JoQg7yqW6habQ/wOZ39jcYNkWKTiuw++JXTDFBADDbKDUPGRnwNOoqhcLP5c44OANbac\n",
              "Lxj/kh4TskZVu9pwq4SjRvt60mzeWy9I1CEVT/w6qRkVx7x9qtH7X43MCJQRM76T1QTidha6HqcN\n",
              "cyDNmQAAAEsBnr90Q/8AE1jM5tZUJjBYlpCp67wWZl+C9jbGPX9GG6+zRHSbBk5d9ta5V22D2rNE\n",
              "z5wc7VRGMQ8G2NfXphR4z1NKJ5wBWlua2FwAAAB4AZ6hakP/ABNXfUDg248Yi4u++sfOUiYouJic\n",
              "VXks1ZcesjcAcTHLHc0ACawjXfxyfjFNhoMVKJLmXAJYaBHJXG2n4IrTdyWQxh6xMtv9iml86Cvq\n",
              "E82EIhpk0ktf3V/KqdaCZcOHupOyDXZr32v3riOON1wpAqufAAADOkGapkmoQWyZTAgp//7WjLAB\n",
              "ya7kKuAE5MArTYEnojyROsag0cKlgT8MtGvHpsjCVhX2inNcsNREIGK/vNUfuYkR40bHHGjMKEXe\n",
              "P80bEe6G9O9cjHNFRR9NuBSUe7uyhb+ojMqaT5rLbcBXFIMf2VebZk6bGoHMPz9u0Nz+0hKpgey6\n",
              "zn3rQ5KnLDIK3kvAy4g1kurmyZRUrHjNhDyyAo9iCCvE167EGqPHWaC6tDr50IkdPtXmPUYedDn4\n",
              "pEbiM1VS8ExZQ7Tb/CnF+rkN/Ofq2UG3QTYZrfoJ07ylXKFXz4YQfmL9Erfr8PnLomBGzXpp5mmo\n",
              "41Km0MGmiJ0rMtB1JBL4fhH0ofPcaUw/nzSpT5BE4Tu9ByB4dqJ/WEno2VKXNK0MnY/Ogj5PXx99\n",
              "j+LXmEDJo8RwF5YuzJCES+soW6ELZtVSAPlLY3rQBukwmeR2o9q9AsM5mO+zKMdEfPWresQ7ia2u\n",
              "4hv9EE/2QFOlK7kCOSPdPdH3NFZM2JIllg/A8fElMXlpm9kV1hnApCy3P9dbBF4CNBfTUDbhBpif\n",
              "qLsfST8QuZGwqRev+MSiTZB8UCs3BWiYspy/PBaHO3aJ+1BEmCQUJ8OPpa6LO1eKe9JhIkCZ79Oy\n",
              "Z0/pspezbYy0SJ6iMJnSCJqoSmx4lPN+7lsFKtJtyncBN5qLhL4rJOJiFhzRgGOKxuWkW3FYlDgR\n",
              "7Kxgp8zuS/1pUIkfN8NcqgxEaERi/GgyOCW0pPBY09EG6dpQ6AO+2Mz3qStdIkHJhuhv6ENpaG/M\n",
              "58uQlt/WKzuaJwOQJlGEuYjO8Jlw3sqllQi1m+BsFRdjLzXBae7nv5TyTJM0+2KwNwVFJwt4om54\n",
              "XNOnCBLjU7EkkWFcW7hLlVLCMn0Q5bANKJF0v0bQEas/mPXaWHP5ZfzC/9KpyzORA3FIJbZMMdvZ\n",
              "Q/I5zuXAkPEoEwBgQfq0MNOvINbcrkuCGrjoaxBkGAjSZXcDsQUBfribEguJRS9/sc4JzMPDOogZ\n",
              "Dry7gRvp68pSPvItGUWDfZMqjv+8Vb1XAWok+AZd5cEH/u4gvjCbB8TZUNadPGyWOgOJLSCmtTWu\n",
              "5HsJXwczQZNnIZEAAAB6QZ7ERRUsEP8ACKOakbuB42ZKNmLX8DMq6frU+JqPu32VcaLZthSXIlIQ\n",
              "SxnB+GkdauUeeADVA0wNpN6T3M/QRJxym36itVKK+S1GBPm7vqX2S75LtLY9O+NbqfwvP4Q7/HcO\n",
              "Rca0t5dXSte0wjHUkLG3e5J8KnBB7ikAAABMAZ7jdEP/ABNZd0XgBxl9Kc89z8Ki7G8KpGppPvsQ\n",
              "6umsuuhB/sGzdnqHB/1z/KaT+ULlfR94BiK9GZ/K9sciRI/9uDTAbI40gUgvoQAAAC8BnuVqQ/8A\n",
              "E1eFf4pXSquXDbZ+nlw1q3/MPr1DUzTsCgG7nu939edo5bUIrADjgQAAAvpBmupJqEFsmUwIKf/+\n",
              "1oywAcm1hkeRuwAczefKEcRuBuLviTG/lgD+KEf+Ra3XJ7ObX1CFa3m5Zm6LF6TPv2ZC9KeKc81H\n",
              "2qYc+8bSCwMA45BdTGjDngZ+DMwol9zXo2TosCte+NkcygOqG7QgDcuESuOqh5mtwRFSeAEsXzCu\n",
              "4PujXB/osPiHYwnopLYk9FkdqypVD9mtxq2+UISgdOpPEwfjU4+E3u57Eha+F1xGFNaxpUDnErYQ\n",
              "wc6HXPTfwXI3bqBbTO1FfAV249WtsVh9AtG0amgVgHdS3TkEhb+ybkMXvMaGhMPQqJX7l/ZYalJL\n",
              "ZbCu/f7RvwyQcyr6VCZDM7BGKKD/jqC0OhHnD/MjmwiSDZMTwKX5DCaKlDMrBiliCG34qQIAf7xp\n",
              "DmzPXyVIJbEFyhbRmpirorGadfbH/Hq3EjbVvh5suR7Yzk0Nad3+13y3qnL6WEyfo+tp0aSfX/00\n",
              "2WJR5Gx9ICEt2RekNsonw+Lwjrb5dcVYDQE884AUnnfRFXQa2rqEEqOMghx/kg4huX1Wmjz4JIIR\n",
              "go0c+coNPFEio4nvvYm6EEedFgut2bqyrMON5ZfcO1ZFy7vCX0R0+zA55JZshvGffSj/Oa3FQx8U\n",
              "U7gs90t7AbW5kw0aKvcNgOhac+3DWr72ybWz8VOVImcBm8UurdTBd4YmVaLfx66Ha91mHIL42arO\n",
              "/tv5x9g22voRmJUb0yKkgAuXil5Nrmc7D4w6frijOWm64LcCw5kj8lJskIJTDNj/gXevuV/BCAuT\n",
              "++U9Yybgs5WpMO32FBa+qWiG8x9y/ycD780ClwrE4QEMtIeV+BDnGqlu/u3NJZ5Gevwm3vsdoOUQ\n",
              "PcF58jXuMwq/KeXMr7axKME1M2zh1bz5smfH/tAp9qKflpOT3LyHFRfgfkfCQI3teK/XNlfyPb0z\n",
              "nzvjjCC46Su+UeMiVl1ZsxS7SGDuHFgFEpFxw/Y9ihKiA/9uYXfMJQFqo334LpmkCuZZk3XirnQV\n",
              "/ermQ308vREAAABpQZ8IRRUsEP8ACKsCm1pf8QgAL2bZt4HEyUMNj83I2a0SoQn3fQU1+r4WGoWE\n",
              "QuIZuywVux5S1v9ZqI77TZeOyexN6maZgdBu9Sh/iX8H88dRBkVBXx/0o0KmghEy6EWhpV5WAHl4\n",
              "kIqAAAAAPQGfJ3RD/wATVDVvL2mAjSAUnhnkLeYW/GMg4J33i98xQKmQEq4p9M4kuc0C9gOebu/B\n",
              "PgEfwuhJPzIoevAAAABgAZ8pakP/ABNQY3QOF+W9D+seoa7NbjEsq4lkALZdQIFHBECPUx9yqmp9\n",
              "HL4iztlovip98iw6FUZMbOZVF7HcWQCelu8rSAyM8faxGZO4NkxQo7TUghVQhr6dvetsIaa/AAAC\n",
              "2UGbLkmoQWyZTAgp//7WjLABx465S7ABXCOhN2YoG7hteD57mmFll/gY1hEQ+zN8vlw4+cVYshA1\n",
              "g2DsXrmR0Wa2WaPxSCPuLHF+QQ7Qt5oOP6eum0Qbp7ZWsXy6S+WLoDMCkEHGUGtP407N5d7CcFic\n",
              "NqEHsF+bQFZOr7oeTIcRBdM42hMaNFgoWxD/az/ywcERSa3Z1tAqrRTXC7DZ+GVsQgYWNwdlOAIa\n",
              "/K4Ko9I+AO0kUJ9oDxGX09jEx5HNX5Sllx75jxLp5glx+jhOphLuCpcuL7z+HUxIc3FwRj91yYql\n",
              "90gY6CRlTRWP8gne0D31oEKeIADlOoQHMUjzOMeBBT6r/5/6yEUN5nugCI+GpUGBnQs6ZbkLwugZ\n",
              "oSVOeK/940cyUFMUqA9tb7L02yCDxrG8x1prh4W6Oo+PNgTh9ufnahqUoof5iz7RvYmnd5sCDwy+\n",
              "LxyWta/d5xhLUP2BbeMnVpziPLxmfjhXP7DkaOqyz8Izwt5RBOabTuW49dDHMOl/BvNmTbnnGCHI\n",
              "8y60d62R/xleBPfYLcazfpQ/3TqEDukGgsMU2dnatyjk+XkoCnEFQmuuGJE+USWpDmH8h2KFmaqB\n",
              "eCLwOdYDQkLyNc6eJ01pmRo+S+5ggNTszObNzJntmB03RQikY7ukp249pBU5S7UIKxCqx0rGXPTu\n",
              "nV+mEonTmP+stj9/z1EyoSNFko7kf836y0KWcQGhwAnqNuh4RmuXS3vvXUKaDTByqUw71Ry1OkCE\n",
              "0GqvJrqdWMomd+XFxneZ5NGcmeD6lVos/GFRIELU4SMQfS8rMXWXduXXUrnsy/iVE5LAulptXNdK\n",
              "QLKCErGq7z42TR6Fq2sGYM4L+JP2MwHkyCwmXRwlnof0L1wWWEEOadw3NVb7rdDxvx15qXAbs9/g\n",
              "Fspt4TfEKIOYhMVyZzQk4GwbIJahD6/GeOChGHecaULtqdDDPqKWvSHzyEmwNAAAAHBBn0xFFSwQ\n",
              "/wAIqoIGS83FdAEMOE3phHz3emDlLz50hC7aSY4nNf5APPntrMA0cJyUfIzp9kTVsUZQU6E5tBRI\n",
              "/zKh9RSyEwdY0Jp3+LtWHLkoxTE6Yi8n6hl61FAbmJ/Bt9zEeSA+OW+mB/wKS6K2AAAAOwGfa3RD\n",
              "/wATVKdHhPpYgtnSw+CR/CWaaAdNBeef7bWQDYtQli8+wRE1pRZg/61l2qA/iPp+/Hi78toRAAAA\n",
              "MgGfbWpD/wATWItcFSbb0u9TUqBkafWNeqyGezdN6wh19T5V9rRnDOHVa6qUZkYBIQk5AAAC+0Gb\n",
              "ckmoQWyZTAgp//7WjLAByNcctiAB7OwOjRJ8b2ouGkcL4B0O9ZXXIksd2rW2QkRQNhBJjYL0uloj\n",
              "sAikg1RDONryk1oW41ipXyH27rG5cuqtxDz68pSomPt10rdvh44UUcJyrV+4tdrNZoCwAf8Q7CiJ\n",
              "LynLrmdRwWs8lCLIfMErg2yxg+oyUcqYhNI/QdpqmG+pB7U5FAFkjcrGTaMF4z15Ttx9r4+dJxMA\n",
              "LwZphYEx1X23z7myVvlpxMcj3GkBeDEZkJFTG8bC7fWudLX29KaGqmY4xNglocokF/wfdlPpbVcI\n",
              "K20Dqarf2nLYXtN21bXkNvB4Xsij3wisvVFOJNXTiw95Z5cam7r1oU6d6D0qwQvLQSUgrvqKO9Jy\n",
              "QSsUMGO87EoxKycAEmMJ0ZIFDXftzWdu/ufe1EIbrNp5yNqq1EWwaQL6hDB5AAtDNQamHgq2zHpr\n",
              "9So95D6qariNRzJCb0ZWKOD3DvXsMNDVhPRZurwr2S2cdNuf+omXYX5BfIX5rZyPXv5VOkrjHz3K\n",
              "C4CUFA6JppHeb0oxUkG2pB65Ug/6ftoZuPgtVYfmW3Sln7nKVV47nYR9ZfvcS2r+jVwKmAhA5Yr3\n",
              "bPbxtBMmLNy98O2qNYUGjZ8/YO58owy/0QeMdMqobiXNM97JOdW9ndrZLMVYvZR8ron+rQaFU3ed\n",
              "3Dod3OovAIOsVIKij95j6eAK3Xthx7rshhWO2Wg7SEOdBOQ6vkVQ33LjyogKGGUB1wZ7iUm2TUKR\n",
              "z93CmAJWSskiYkNF1++CnHVTx+BxMGJ1R1iiNiMOIgtaNOWm/C7KRFu//8/l4SWRRmZFozlajk2O\n",
              "Y/HDxL4U+imCFQLYFBHllhYkYA0x6Vg42afF9Ly2Kcj72NEH9OJS3lzt1q9hyJDzS0Ok9Vwt9iqj\n",
              "K/MR8oymoPfKpu2GIsM6TB91K7XbASCIdv/CNsXNyyBvXVtT/I2d2dTWu3zBQ+IOrhb7zGWhs/Xh\n",
              "UlMcgAAQf6TEuR1aosEnNo6r9mEAAABeQZ+QRRUsEP8ACJ8wk+tC4zyalWNFsbIGe+1astppm3iS\n",
              "YABamFfN4omMQMxYknr/qH3gRC9cqKziSOMbvALCn6L+fD/YQyCsSxgIiFPoGA1+FbYeJs9N3m66\n",
              "RoMDBgAAAEQBn690Q/8AE1lxEYiVWEoARbjBzH8+KJcNDR2BYRaWA7IkBJJMNsXAiP4PFJd/10c9\n",
              "P6UbCKfh/ndjfxAfLe4+G73LUAAAAEQBn7FqQ/8AE1iLXBUnJuvOc5o7eM60xgMpWGmJDWBpCa8B\n",
              "JD3pM2GZmdVB98KVLPAagHX08lXJMGFKBk2PSLkbIdD/gQAAAvpBm7ZJqEFsmUwIKf/+1oywAcfc\n",
              "CnNtgAgnxTa48ijceGeZMX1aku+TwU3E0uw3shiHpTXortSYQ7UjDXu05964+IA1IFHlC3NKA8ma\n",
              "hdmOEK+2MIU8bRat2s8Glrmtjeabh682r220Kb2LmoaUmSzqi7NI9bMfMqp402bwGZS8PEoaOc9Z\n",
              "tnF+bVa374MpY+Ho7g/yxYze3Xt7Qt/mc+xr6OyVx1ffnlh4gZ5MC5UUWT2lPPGhc9ibTJGsnwzh\n",
              "+ZcqfvBWpC6sPUk68QL/w+SgGWFh9uVvKGkelS7TFQ0pNi3MjziXikHeCUAcV8zLKWUzX0IFVq//\n",
              "uQNG6S3evJHjM4XedZYEpg4oKfWXHGH3A6SxoIgFJ9iiwxMLMFI5kymqIBHpdL4tx33pbbcvKAXh\n",
              "nm61pXURVIyO5k5WGKilpwCCYgHLLppbOW8gd6WJvDCzWwMv4b172+WiY3zNxjn2ApiDPlnqGnhW\n",
              "fKhT+jvTctP2ZU0zs74XrA2BNtvI8DR25n6j0KXAxk9NFqF4/ZoqqgPmzZtzeBTSS3wf3pHnnK4t\n",
              "vUMk3Lw5ZncLMrezxYKudBQ7IgisJ4rPY4umDyVaufoIZ0p2VXfvBrZH6/6t/cRTmRoo1NQGIlwG\n",
              "CwFmAvppxt+PQuYlECNof4pI5pxmBQV1PYFibmT2mhtWkXl7cx81BbKpH/3wJiyjH+CNihHTAeMc\n",
              "RY+OuH2vM1/hvy3XCLpXLH+hofzObgFHvNs/g2ci3xiImqQC/I7/qJXYWQxiedUaroD/127ZFDWL\n",
              "QRx4itXzklzBx71ImQZ4G2vNvGcHDkQh3q3qKbuGAQCvhoAJu0IAqrkP7aT0lGH6S1v9hrutbUu+\n",
              "N4+wvWAN2uAd8r4+8Y7BiCB4x/UB+QZo6Lyh0/nqFNp0jmmv0bsSPvev076Vx94it4ga86a6QthA\n",
              "i8/SA0wkpHsnslwnezXrwwh9ANma+Nnhf2UY7EMr0yEwyGQYbtugbhuIesIrmKNmiFE8/NgV2Py3\n",
              "zoAAAABnQZ/URRUsEP8ACKY2Jkkk/Sf2s6ROdOB6rg1XBqSlV3xkVp/8pY0RAVVABw7Yv9wHQmbU\n",
              "+SsHgjL+F3RpJXhTGPPJGUMlxl/5Bs4qvtgVottKESEzgjiig1BZL1UVNglL4bn4Ug5mQAAAAEcB\n",
              "n/N0Q/8AEtXK+ZJgKkxIRiqB1g212Auzbd5f3VNFZ2d3LOABO0ITORcxrfsx6YDXmQlC3/0OaRMf\n",
              "LMHh4gkLXJ3cBoQ6IQAAAC0Bn/VqQ/8AE1eO8WmYoWB6uGy1vN5gE0QojOketybInTRD/qxccAAi\n",
              "5juqQfEAAAKsQZv6SahBbJlMCCn//taMsAORrq261MlTj+RQADaGXjOJNTq1Jmjy+M/MRNL2b1dc\n",
              "42IOa35wmrhTqg2gujyi47rlyxPNrXafS1LV0AhazoLfKQK0ip5dniBZuzLlCdpOdKtIqZMF5WTT\n",
              "KPkPNsupSXXg6QxutMm0yjpUIbp6fvKLCOM9wnD1aOWPW6WP46bGtsmdtmVRp00hFCddaUs9TfnC\n",
              "EnF6hh4QU3WAatO9aXp5UYrYwZK7Epg50yr1vvHwZ1fGX7av6FSBCCR74LLSsyvyfZMDaMT9nq3I\n",
              "6Kh+eoyZSAJKfutx6T65vJf+AM37PkFVCst9TVZa+MJi0T7cCx98zcqfVg30MKL4Zm1DUD4lBzJi\n",
              "HNoJ/9lPC5hZS9zEMWlAhVH1ISma7/07Hyoc/WdhIPVF75MOfHjky3MLWVzuCWOe+AgIf45rgv3q\n",
              "6yW56jWGxP7BK9Ag5ZcieqV1EmajiArmd0dOPevFDwoXZqHLSdWOZnxU+4W01Wm8Q8v7zB8Pcs92\n",
              "xujAs9x8DxeLu+9XUabY+mLJsu6Lf189K2OHcFnsUGlsgXZVhoMD8YDRA3augYo2tQAlceN5oUdQ\n",
              "D4M0WBQSQm9E4mJmKdCM5Y4S5cHebnJtoFKmTxknxR6vCglWcaXPHnhtbNccWR88YZtKScXQF7EY\n",
              "YpbiGCMilVqb184zRg51tIsTV7n/3/fmR/jE89TSikr9mJThlwFdMJSl08vNUp0jXrwPWaVzQiAQ\n",
              "E+rvELWQc2SGS1OFOC2pfSIklSDbKEnPEXxf9cVuoYjB524Y7GMtnnB3wcWX2dk4g6eZc5jgvque\n",
              "ggnP8l3lnzLIoO9NKe7OgoFfoy4IWevJmkYU+9tTy4kGf+/0V1qbkPSKrZOFwSFxfKbdZ8ZMTLTS\n",
              "uWb0ufYFdSoFAAAAY0GeGEUVLBD/ABFWYreCN+MTANtBh1Bz6onhWeWy/bwFhAhF2AgBZ0EevJIj\n",
              "loMjzfYQ5itBWz+gAgcgH0Yt67TChjhY7Np8k2lPfST40fDMqhdZl6ushl6RzbvyLnQutRcKSQAA\n",
              "ADoBnjd0Q/8AJrLtlRDqGwKZtc8MqW1uJIrPx30iXzj/SOHncXKNLweSMigd/T68C8KgD9B/SZEC\n",
              "ASG7AAAASgGeOWpD/wAmrqToj93vdS1pFzWRHdOFfDmhj/g4ssnXEtP4nQKfD7aIQp2bMAHWMVTN\n",
              "qQipySqhVnPAVLEcHQrXhR1sONTiAMFZAAACz0GaPkmoQWyZTAgp//7WjLADka44juABmpJdvnBY\n",
              "CouEGMWaGAjxybDhkjQ+GM625V7mPeOE3RaCnAzG76sQvmKKSOys4TU/ITZ348/4bX4uEJc+/qvo\n",
              "zmF8f9S7I0rC5nfxY594ijdVGAEYYvmz5+LSmRQZ1gr4K0nNqiWyZ8aH9fvVFdzoIDr0Ea2uqVU/\n",
              "Nj+3YHY86Q+5sOV/aE7CUnb2HYgC+QtIsLN7JxovxES6Q4fvgWJSYRx2pXPTNul/hVGX+/S9LpwE\n",
              "wdH2D58cvyaiS/ndmX6joApntnoaffgMwFAPs0c7Wb2zKbO843MVNhKOrAk4uYE2lQwQ5O6DYjla\n",
              "vLgw9CXlBoNYCkhLoFGHkUyP8td9FxAD9Tz8KJaslHRPxFWoZfNicEQLy63iZQ/9wZpC1C5pbV92\n",
              "fhAX+IhwHqM6B0DFGspKCeKTkjKHDjkgDtG7u8cOIAior32X4k4C0BzJ2AVPr2VJXkBfn4xwn/iI\n",
              "rjLrXTa9mWBM43ed92BLfoDqyqSYFZN2aS5l/L9kOKyyxif8HxXdZIwY/7DAJdjfqalddVfJccZ7\n",
              "31gVjWz6SocX42nsLjzW/2BFxyYTQWew7VqVgBtBj+6yIqAvfC3G5HiLUjT7aw7YPLITnhOe0dUb\n",
              "MEpqhmCx/vgg+wBNScPdOsVhxUgCTHoYSH3+A6gwSj4b0F0XDXARYq2HIyv2EItLlJNBd/F/Avg8\n",
              "hfXx3O9w51IASZ5VJQdiIpBcYeTPiw0adVT9HgJHTUgAh3PyabLHfzRbf1YL7ZozrwbrRKmDb+P+\n",
              "RAd32m+FupzWfzhFI+TLxMqxL5E/PNVT7lVt5q0gqOZvklYw2HzJ1qGoDAwvBX+nUu4j/15V03Ia\n",
              "z6IRy1qOMjfvs8a3qYvBESzy9rA+frVXNQm7u/8IMKKHo8dwvKq2fVdAuyxcTxaSQvb6f+hFedCH\n",
              "xC/HBo0gAAAAZ0GeXEUVLBD/ABFWB324mH4k7pJudTNzbQlsf/Udb2Z7+T231T1s6C11PCr6I5YA\n",
              "Ek3pIbYyfb+kXB5pHkXfxnkcjt4Y4eJ6aUgjWzp/yaZTl3XRuQyH2eUfGQ12IeBz2pCKJjb6IeEA\n",
              "AAA5AZ57dEP/ACavHOeP+9OYEt1mBrm83nnfzMeIfPV8a+7HUpRkqS+DWzMlb/CMVvzP+ASQHL5h\n",
              "ndEjAAAAOAGefWpD/wAmsREdRHqGwhJE4zpze45l9HNak9V10MecCZ8q6uSdgZx5jiJKrrir6YmM\n",
              "lwGa88GAAAACz0GaYkmoQWyZTAgp//7WjLADk124QcwfiHJAAp1jgsla99y3OwS/XrSzzEHptQDh\n",
              "mNvXi/DAh/idMaKaSrNGAPwwGIpgpwNRssPX3tukM9uemL9tVTTeZfLMWGKO4u8EfK4a96eAOPmx\n",
              "6Pa3c6SxboF9qfu+Kdo8v0u5lXNkDIP6j6Y6rmfn1nnzbmh0DDLt+9c21wFQ4KLgGN3vli36Eq9K\n",
              "desz6VwpntPIjJjsHfNmo1xs5ZDZlgQOsfuXS2afALFOTcNYWGn0OiERFYOF+9MPdenF/8KY44SS\n",
              "RaVjnEts5ygbP4UFM0X/sT9ete9wTuZfv7OIM7fEc9pcpC1UHgKtISpefzPFt5m2zAdIBvkW4hwd\n",
              "GwJrUCFWS3NpptrgnmLi9lKEAz08IiemXYV8b6DG2IGzNdLxAlUzgDWM1csrXOI9N8Av/SOHhTHo\n",
              "hGpzeN8yWNxeQ3s5iBvxgMBdbt0SR9GsX7Ww7HbVAG2mNySKRjlunaFqMYpMqH/11tnP2Jl4ZkfN\n",
              "M4KBkNU/s5AqE898rhxkAYRHCaso9CducXdFXnAXgYr+wkABgqa0Oj3JROkHzYos9QIbVuttWGpB\n",
              "CdHOiCvmYu1ojW0GNQXLCy0IMcBYwwi8xGxpe7XfQZFBbv1Xyeob6hV0n1O+Kus8xoetcAbM2d4t\n",
              "aUvRxkdo9bPPzN6rFpvDzoOCWW88q3zpZwwO1kNUt9MoMUtFk+qQv4wTEDYlBnI2GsA+o4gstLi6\n",
              "6Mwj21vGjg5sHuOeZ3RqkDlGR6s8lWaP1yMqttNOUBIxu1rM5atqfzdP4YGKDrdhrjyimfffrG4t\n",
              "yrzphmHPqYhvu2ba4hQh0p9fhiOmaM5fYsVTPN53/iCPhnfcHdifYh1eetmb0yBw5KdzjnsjVvyU\n",
              "f4vdanmmU12NQ8cpr1CV9RzHfZHK+V4ZTWvPlLpSGN2Oz79byoDYQ5PjHoqAAAAAXkGegEUVLBD/\n",
              "ABFVBA/ZFKPbFqmF1p0aTYqQ7j8y0jA/fzeqTUYUiRnsJmW+zmRhThGwBo+M8u8V84qFeWQhUphF\n",
              "OYQJGLOPZQWQmxfADKu/mbOf/Ltx0xU/RUouz5kAAAAxAZ6/dEP/ACay7ZURDjbAo8Eu/TsI/8bg\n",
              "K2SzxEARAbujH+p57+0Cemy+XzfoS0wZ8AAAADoBnqFqQ/8AEtg2jvdc357N7CXIxt08ByfdnZ7y\n",
              "7FO8hZJHKc0YGBj/GHVehaBSJTt8LH8okC3j1MHLAAACvkGapkmoQWyZTAgp//7WjLADk2qSvx4g\n",
              "BHKPnN48MzA4Z9CNgooHyNI1ujqFbddV8wP1smPodxgLJTZleB3EC9nvEFZ0NFEHlA4nR/u8s572\n",
              "v3fWc4mV/s9xeraxrROYO1NLXqs6bcQft8f/bM8nZGNoh9GVZKfFE5g8/nM/qxNCU12yRq+04cQa\n",
              "bTYi0ve9j0c5813l07evbMdet1aE7fp5WmJdNKUpNESeatb76l1N8XO+K9g9bQDsXUWqgO897kPN\n",
              "FFLftahNftCsNOlHHxJSIFKuk+lkfk76Rq4Gwt0RXsou9Y6CHVXji+zIkO63ntP9u0fXyZ2S2V5/\n",
              "RZEQBrYbQp/lShTxc+srcDEAmpLCgG8u9IZFIFrgBH7fNqg5zXwoWC/j6sgbYg9pdGf0dZpOYg5q\n",
              "afQJdPjngcVFd4J3iMtocalxE6n66dp8+gOGV+EIM41p/Q6wDco1581YJ8HZSAtxBPyYSuXZfU0v\n",
              "FrZ2e1WEXceOl94mkvCvL43yymzvKtBP/aZcn1vVSBCsCIhMHM3EFNCkRu0GiM78r1Eb05f1QOCt\n",
              "xu1xGUx5c0vkWidUZSJzzsxPSi4+JLz+Zt1fxkTZsgEyLu0MzzWf4oRbuT/ws+/vOvZ8wTDhv2HO\n",
              "t3j4HcYN5dBDM3dONLctGyeSLbywTQ5OQoofgMj0CAVf1YqI5PEtogD/lDATRNjqluAnn2TF3/kn\n",
              "oqljKwabFyKSlwdRP+4sp3wuJMlWFLX+9HbdnvG/cYvnmSJ0TpcLuFLpbhqWnh7GYwyQ/TT2Kj1m\n",
              "taJSZd5cbLC2qCzTNT8Gb2SqFOft0qXjs6oBG+G2LbYyj3lI+6KIhf1syYc+KTu5xo7bat0D6wH1\n",
              "X5/8J7ip89BrcS4+Dj3i6mLmeQtww/d/NqdBTff/C4SOJ3Kua1Y2eA/UxocjXQ8/X6FlwAAAAGtB\n",
              "nsRFFSwQ/wARSZwIV0KYod4VZGGWkWhf5QnJjSsAUnOG21UjhdDHRPD54VRoMEp+ln9LhP/WwAXB\n",
              "cOsZ0wY2GkTgDUvepX0ymJKk8EQrPLoj2Yoe9UdPgDiU6YmNMyYWGgSZJju8oYSGtQAAADABnuN0\n",
              "Q/8AJrLtkiSSD7W9MLTCc0sxIVaAbW97mhyvSLbDNEJa/IIOGOEGMEpw1YEAAABGAZ7lakP/ACav\n",
              "Cv8UrutMgM0XZVCZ+6/oVYp9N1PDhA5ZWRyNEXb4xEAA1GSwl8YkOTHF3q1QSiYh0Y9pAxB6NawL\n",
              "lLUKWQAAAk5BmupJqEFsmUwIKf/+1oywA5Nq0sDINAETZXfDBaKpwzQ8fVy3TB3NXKE9VBE9lrXh\n",
              "BkAgFNOOUl17HfAoh+IT1/dhWJVNalPvJpfpKVoktbnvEPu/TfB9DBXmGqG1ZSw/22rzPYzydyVo\n",
              "UIfdgyE0L7HjkeDttt/nUP00VsYlqahkhMM+8UGOsHr7blDryACAqT6c3n0/Ss/O2eXjnVKGB80Z\n",
              "Tfr3uPyBsLKLv9GJuINvuXVhCpcnLUG1gM2xcZ9JJE64qCLk/SgF7onFWF6NTVJ6rQcMYxPt1Ejl\n",
              "AEsDzW/fhNtBbyDUJL90vdEamufC6n5iLDwo2CoDgcFG7CF2YUWuj3VFv91IBP27LR5u6vezzioY\n",
              "t70zZ35wgiEnPhA2T786CPGHiD4RWmldHb5GQ0udbx+7taLgHTJYxj/acEEQi5QRnSYEom07s63F\n",
              "0/cJP0LPKHRiEIo+Dk7bBNLUUUdalr3up+c67umlDkdehML1hSuEKhTJDhhZC4YcnyAOKJbzYlVa\n",
              "bjFNL3aANt9De92eZKzALJuNrKLj5N2HDlG/gOobM+7Hxn8NCMlnZ+OszjzAFQVqPmzEjxsMkYMi\n",
              "7OUpSJnmE4ykkUxCdkELOxIUrO8Xgjm1l44+2Vx7iNZ1sBbOXfoFiYAqUwa1LBCaaVO0Y1J8pk3d\n",
              "g1sL9JatcKkZt/MDxnVjD4IOTyxNdDNCiWwtQw4VL1g9GV4lA+dew0BZHKl94oVx+OADPYTSUka0\n",
              "+tryMWVEa9cEdCqRjEQrJG+PNgv0jdM/4w88wQAAAHVBnwhFFSwQ/wARUqX4odkzZD/HrrsMjfYg\n",
              "xSDwcGAUG/kR5a+j2QcllLTfDwauxayRSl8qrYYxFB/OZQP3J+ds9JvedHMYlAQzzc5XkAQNTnpt\n",
              "PS847wd3y2z7fc+fB5P9yeHRp9e3FELKFXh1ahmxDDDnB+QAAABYAZ8ndEP/ACarleHr8tR1SPZ3\n",
              "RtbyuPbV2iHnq9Tf0h7NNgoD8UPax6U+qP7T8UnaRvHFYIB5ldTSK7nOmnOIES7LtovpaKZyK9Ws\n",
              "O6yqF9YqZGgXIHof4AAAAEoBnylqQ/8AJq7+R+/3E8g699vT5SwJKxhpbX1rWdvKofjBZr9fva9b\n",
              "xBdNd4pIdF66ancT+pyBQ59dgjxzqX1I8B9JrX8ZYnbi8QAAAh5Bmy5JqEFsmUwIJ//+tSqADecx\n",
              "jP3ABECgU0PuHR92+OsgrwMwmHtOxSq0XGodaci9MM4I7ugVCmrjPe5neELlwOduS5PQG9KYaUZc\n",
              "PM4V4/3XA5ON2a+mmizOTaIyX4tjksSt3sxp9Jbt7nU0hxKzFIbF/e80/LOc7TqpkLwirgIzuEe0\n",
              "XlGHqpVyTr/jmPjwlP6vDIcfU1qM1blxnW4rUujWGUID1/lN62ZpAS1PX311RaseyeTm0nJZDwZo\n",
              "euF7EEE67uvM30GzaoOIRdqwaGUdxXGdItwo+o2j49HLibQA3A0KVUPCVx5fbTgCRpDaC70ipzME\n",
              "UYMeu+sE4RQFd/I/9adlewFqn+r0xQO6C3ZGWzJs9hmBXawcn9389YPNSTSqAmKlReDmHn/Nan0r\n",
              "pjETSs2hNuZ32U4+HQsZ2qhatbtsdglpgROvWaQUOgYuUI9j3Ko4YJgAWx9oaOfvpGgEWgZfN7qP\n",
              "zwx9J1FU+MuxuF6u2ZU6XacS22uQfv6EGIFC5K1TpcLz9fEOg0+IP4sZ0SczPw2ykVZWdjp0COWI\n",
              "pgsY40ryjWtBXosPI84TwwbeGkOh+8UBd2BuDKkcOlyo/qtbjdRhSSwe5lYy1yClDQvuYJfnYab6\n",
              "z/ihx0YnnnJrPNmDZlpQMvraow6N0YTE3ZZxcOHjNtL3/pQ7J7UeQz9R0L52BoO0AcA/6wbETURV\n",
              "K6UloriIBVgWBAAAAE1Bn0xFFSwQ/wARUqX6BSGbIf5RqtjJ4usUumuwZMAnkRbu4AH4eX6+NHvC\n",
              "uZ+8/r5+m/3fMFBG7bFYZ4gY2kokwOLV0zUyoO/qCGkrMAAAAGQBn2t0Q/8AJqlOqkHBORrWFgUO\n",
              "nQHwvgmZfVwI98vsVxYedkTgVcqLdWbmCo6frg52Owid5MKC+UT56ACCwriDvcKrxlV9jciIBSYB\n",
              "V0/3Q21bqbAtoaHsMpYcQjjP45MyAIPvAAAAZAGfbWpD/wAmrwrob62ddegm7aYD0uTB8fe9MNmI\n",
              "yJBjWOt7+nNwYokJnLFrQv4dpYARpFdv0pBLFuxn7WM+LhuiDttoO0bUq9PRb9tKHnedn/sjSqAJ\n",
              "jSJJuVmvU4mADUvghs0AAAH1QZtySahBbJlMCCf//rUqgCESbckuACIckYUmvJmrrfLcwoRy1IiX\n",
              "0aVUi1gVHByaJLQLRo7Hj2k3dzvRlt3uzhpT2SNqw831eNLLpTeS3ZFGw88Qu3VwCkRpncxIhYa4\n",
              "IJA9GMwFvU5wOt7lJgEVORNqoS/muLT5Xor1S+lSk4E9yUrFidmASCW8rTLoN8leF7uFafKz1Dko\n",
              "1s6Gpbg6l620nAggmBV1rg7BsgN/pz1NWkuvmN4XSo6SnRiEeYpQUWDA4r6J7/mpVG13yu7lM34S\n",
              "fRWhPa85gubuNicBP+SGwShvaKCxZqkQXFeizKE5FlbhJnx3NQjtAtV1/HWyo4XbbedxiomSdG2g\n",
              "7rH9Jvoky+Xa6cOWiVLEyz5l9HcBbEAIXyaHSmfwb6GmJmNGNldXS4nZX4MI8FYmMpJMva0xy4Tr\n",
              "CgRrilyRsgObDLBpC6I8ZM/nLrqt2j/rjFwEzpkZuvJU/XDSZH97rGfE2DvUw5BZPx0BBJTFBmO1\n",
              "uarylDcDTNFusrZi8ICq2M6Z/1fQxDiLDaJKwLEjZ2nx+orVTjtl7+1Gbnkr5SfzNsIyHH3RAhsn\n",
              "wlhvLZHu9M2C7Dzd8Xw4gU7FYCGNTajqSXz+Z2wBWdAhJa3v+kTBK6dWTlTCVwvf9uYCFGINcsV2\n",
              "QZ7PdRNhAAAATUGfkEUVLBD/ACKa8AA4x5nPcTHtf0lRgRb+dfLAYQ3908ZTPVK/gcBSLepjcAjN\n",
              "S1q29S7id0E39/RofzSPS2/Yb2KFCRFx++SZHwf0AAAAWwGfr3RD/wBM4zPF2C4aSYAn7LI1Qlza\n",
              "KlBrbWxudbIpw8V8uua6AG2y1xHVYBHJq1PO/wlOloBXFPs9Yl8fshCCCjTY5YZ7NmwTpa9u0s57\n",
              "Ik3zA0X8JvJqCBgAAABeAZ+xakP/AEy4ykdwHX8+AIPbZUbDX9AmSHbRum66Jk6DlI/hqSDXd+BE\n",
              "emeoMJWQK7W9JW+Pm2uhPjG6EtjOMRj96MhnQhzIxl9GIcnBzVW8PLAvoNmwWGgfE1/dxwAAAdFB\n",
              "m7VJqEFsmUwIJ//+tSqAIRJ06ZgAibawn6PCb5MQUgC9C1+B7B2JN524ITsJqfO2dEwE8dDeS/ni\n",
              "Z1vfvMaAuLvP2UjSn8aUnjAC8gWwA5fkOKOUR4RPm8r8Bh/ccLkcBbBJho1Q/lUeoNqKUUh/RgOF\n",
              "PCf/RN0btFZyg88uCImX/sZK2sMtesq5uNY5BpxxszE5qVdvUwM22q35Ne/6FALt1o/CYC999+2Z\n",
              "55i3YC6bQXx2Ud9Gc3qOrt3hpP+RRGDl6wELI6GeToENC9VBgWPgGIoiNyTLl98v6MDaVPL0D/6O\n",
              "L+u2T7L78FjF2Dr4/dY+9EHNDtPe2oscS9rwFAJJjLRBdnzBCHhNVBgpa9H/VYdfssHcuAghJcKm\n",
              "vAZt6j5x3cPj/ocJe8xasulQxNMNqHKs4bLko5rtbit93l47Qv5pH/fOLaxThBrskVz1PDAp612l\n",
              "w61D7Dq67ckzicgZxh46MPgnmTSP3HkGIHyPTtln6gX3+e0o4qb6Xrlz2Q6hHQV5kuiVNuuzonSN\n",
              "RJnuhjHwtoFMsGC23neomtUGToGAq65pu+W/2xu76fwvwuyrKVttDGAms9a0EU9XvYu98RYUArtp\n",
              "K6KI7z4JEvwAAABhQZ/TRRUsEP8AENbZObDdFiQW0WsNOL4EunQrKZaY5muzhghX6uq7f/4+y0wc\n",
              "XNQfPEAA4Y/DZ/Qp7L/yEELaU4BlZ1GK1Q8C6GhDnjkjQ3QXODzXwpZ2ccFDFlxskUh+8AAAAEIB\n",
              "n/RqQ/8ATMGSQaxJjmJboeb7nRcSZrHyPNiwGBg4prEoHslqwTJ1Z0Tm/to/1Mkse/SJSoycHweS\n",
              "CyeLtooCFsEAAAFiQZv5SahBbJlMCCf//rUqgCESc84sd/632Ap7sAnuYKSseaD6+ZKtZP84ArtH\n",
              "4uQtaXqLFdCBaAdcMSvHfpsiuH4IBTGQBQIc/m0xDCxKgak8ZRYcVJ2C1fffnWw8DFyIl/zYVOig\n",
              "tgGXtkEhYYReFvDAPyfCCSgv5kbMEgfASweko6FPBmzOtapkYtDNmIW5ljxEj4SC9DIWSPPB3aEx\n",
              "8P5kgwhq9o2e97e8tBJtZZAsNvuyIkUd0v1HkeV/T2oQjDxUtJMK3QjYgPPgOI35IjxhQ0KwWmqr\n",
              "YzqlhWDI3QBwFTO4kWTdBvV+vbdpbOkYA6hb5pplW029o3oib7qw+RNsH20tKidYaW4n4q1yyLVO\n",
              "7lcJfKZPN5uEGGn3QHzL7E0G0my4l7y/Dtez7BkHTdQ5EIKBpUq/jrPBBXeqVJX2p80kXEJI3iiH\n",
              "OUTrnr2qof3AVDT707ZRjkUs/WU32eUNAAAAakGeF0UVLBD/ACJqM+8hDq1+Hfcb31KsNThTavak\n",
              "f2bQLYpUuu6ABtbZZHPbOmAEIVw4SOGMU4kQoPVUsJqBroM6O/gk2FxIyyMYctqp92ZdHqbZsEGA\n",
              "F44dBp5EQylbrzEyCP4fa9W9EUcAAABXAZ42dEP/ACWrlfQa3DV3XTkx0wMFoxGsZl6HeqZmFW0a\n",
              "oO27jhgm/bEXe7Ozdl6a+SaG85tBBAEsSfvFb42Ft+kUvw4gMyUmHvdQkqE1lOG4W8i0hG9BAAAA\n",
              "KwGeOGpD/wAlsG0VpGw60dccc5treFUnJVD/oVkQGytNSjWU6DM7gz9dF3AAAAFcQZo9SahBbJlM\n",
              "CCf//rUqgCESdNxUN03vqDZJyTWH+/rwYXApFuRADZZf3b5qazTNeSEob9MgmXcjGvL3i70JlZzi\n",
              "mKxKt/JEOFibljy8o0u79UAwnYyVBB6aDdk84bDGqK86z5yxZu8+ztLaEbzLOW/i1DgjqlD2nE14\n",
              "RYe1Dw/u9DOl4c/HwurX/H3fryKZEDtmAvWkhUv3Qa0MT0/5xqYfdraDdQCZ56tspa2rhbbDpKdi\n",
              "OkaiubYG2ZYh14M3oh+L+uAtwQhy5lTASHcR28ah6QI9AzXAh2WJfowa8OvZHpc8el2suPkVOXL6\n",
              "cBDztoxgm/8lJ5sQo9Nn+IcVgxvDN7aiqhuIvzs+CE/XCUeGBPRN8m8Yx1Wa6/P2b0GIBDkTmjcZ\n",
              "x460xyVFBG2WwDsyi1ia1SmtMl2s81Jawotx7TWtwNcqsc2thPnzDS+gE0fIP+qwOj5kTqKBAAAA\n",
              "XUGeW0UVLBD/ACJ40OzZxa5MxHOKUdGA492N/6fvAuErnZiovuMwgKIuC/4DX/pTxqyVx/+DABxV\n",
              "yD6eLXYlz1JedrD6G4YgkWoojHaCkE4MjgJ6vaDpfMsLZVAQ0AAAAEABnnp0Q/8AJauU90h+zKw8\n",
              "uo1CTmGwPiqnZ/Fykz2Nw4mmbkWzXLstO9Id5OOpuqDJeDaEYdYlnfiIuDfcE5shAAAALgGefGpD\n",
              "/wAlsG2dCvW33WYwQYHmhSlHEcuznp8nWmme8gXvoRK6702o3WZd2OMAAAEJQZphSahBbJlMCCP/\n",
              "/rUqgCEKbF8TgBLKSvjtSLwK4Is0D7lvaI2PwjsO0cUCJzUG7EgvNF2sWkpCNGezwt9J76Oqzmi0\n",
              "ANIbg6ZF6mCeELdcYLeAva6PX9YztdpBSatMF2IN+M4FwYdMrQ1DJazzy8grTYBh+gw6afWYR0td\n",
              "IsK693oR8uDb89l2uz0uGQ6qWLwAHMvbEy4dIvil+yNrnYgSZ9IxJkQwBICj8f9TwnT3RmQ+7DJt\n",
              "pxTmssaBc1Wytw9AZvKBDBoAeJYghE439Yx1IYnD9motFLMyzwnd3+ugMfrKNG5/jBeMX/uuIdk6\n",
              "ZOpo7Lxngjq/k3xcvyhP3WP2GCw84xcPHgAAAGxBnp9FFSwQ/wAiGURwF4AAJzH3fG/XJA0n9ujR\n",
              "r5pvfiJZqknOjxmVD2p+DoEq70IVrpIaQzYulstteBMCow/CDKghhPD5VjUoPA1vs6/4srWH3S/9\n",
              "nEeoqN6on0bVxRc7U7SJKzTijPL1xUwAAAA2AZ6+dEP/ACWrlfPASlHTc40Vpm1T59ZeO2plQTUl\n",
              "efEf7eue2fmbbW7eLW4zKnbGNFtreEhBAAAAQwGeoGpD/wAmsNSFAMpubUVRr473HqgnQfs40hMc\n",
              "7DdTLjzt6Kb6JPktWGPCAh5ve/ZJg9X5lIBjuHR83g4rWVu8xEAAAAB8QZqjSahBbJlMFEw///6p\n",
              "lgD67nBloAP4D5yYeYwQGFfVw/pBIkrVzRnVwmaJWNlJsyVt/rzx0gF7hMNtgzhtd1J+CU22MLvl\n",
              "ZVvv5zlZXolSjXhPx/aFotvwng6fVMXajCeaEuPeycoHWXaqz29RBbOcA7uAVkxB47IwCwAAAEEB\n",
              "nsJqQ/8AJbK0XgFgStDHdGf9zMuXG62E08jJhesNYSXJxM5ZylRILoK6qlddAgffu3O+3ZrvYMBK\n",
              "ifhEJBCcgAAAB6Ztb292AAAAbG12aGQAAAAAAAAAAAAAAAAAAAPoAABOIAABAAABAAAAAAAAAAAA\n",
              "AAAAAQAAAAAAAAAAAAAAAAAAAAEAAAAAAAAAAAAAAAAAAEAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA\n",
              "AAAAAAAAAAACAAAG0HRyYWsAAABcdGtoZAAAAAMAAAAAAAAAAAAAAAEAAAAAAABOIAAAAAAAAAAA\n",
              "AAAAAAAAAAAAAQAAAAAAAAAAAAAAAAAAAAEAAAAAAAAAAAAAAAAAAEAAAAABsAAAASAAAAAAACRl\n",
              "ZHRzAAAAHGVsc3QAAAAAAAAAAQAATiAAABAAAAEAAAAABkhtZGlhAAAAIG1kaGQAAAAAAAAAAAAA\n",
              "AAAAACgAAAMgAFXEAAAAAAAtaGRscgAAAAAAAAAAdmlkZQAAAAAAAAAAAAAAAFZpZGVvSGFuZGxl\n",
              "cgAAAAXzbWluZgAAABR2bWhkAAAAAQAAAAAAAAAAAAAAJGRpbmYAAAAcZHJlZgAAAAAAAAABAAAA\n",
              "DHVybCAAAAABAAAFs3N0YmwAAACzc3RzZAAAAAAAAAABAAAAo2F2YzEAAAAAAAAAAQAAAAAAAAAA\n",
              "AAAAAAAAAAABsAEgAEgAAABIAAAAAAAAAAEAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA\n",
              "AAAY//8AAAAxYXZjQwFkABX/4QAYZ2QAFazZQbCWhAAAAwAEAAADACg8WLZYAQAGaOvjyyLAAAAA\n",
              "HHV1aWRraEDyXyRPxbo5pRvPAyPzAAAAAAAAABhzdHRzAAAAAAAAAAEAAABkAAAIAAAAABRzdHNz\n",
              "AAAAAAAAAAEAAAABAAAC+GN0dHMAAAAAAAAAXQAAAAQAABAAAAAAAQAAIAAAAAACAAAIAAAAAAEA\n",
              "ABgAAAAAAQAACAAAAAABAAAYAAAAAAEAAAgAAAAAAQAAIAAAAAACAAAIAAAAAAEAACAAAAAAAgAA\n",
              "CAAAAAABAAAYAAAAAAEAAAgAAAAAAQAAKAAAAAABAAAQAAAAAAEAAAAAAAAAAQAACAAAAAABAAAo\n",
              "AAAAAAEAABAAAAAAAQAAAAAAAAABAAAIAAAAAAEAACgAAAAAAQAAEAAAAAABAAAAAAAAAAEAAAgA\n",
              "AAAAAQAAKAAAAAABAAAQAAAAAAEAAAAAAAAAAQAACAAAAAABAAAoAAAAAAEAABAAAAAAAQAAAAAA\n",
              "AAABAAAIAAAAAAEAACgAAAAAAQAAEAAAAAABAAAAAAAAAAEAAAgAAAAAAQAAKAAAAAABAAAQAAAA\n",
              "AAEAAAAAAAAAAQAACAAAAAABAAAoAAAAAAEAABAAAAAAAQAAAAAAAAABAAAIAAAAAAEAACgAAAAA\n",
              "AQAAEAAAAAABAAAAAAAAAAEAAAgAAAAAAQAAKAAAAAABAAAQAAAAAAEAAAAAAAAAAQAACAAAAAAB\n",
              "AAAoAAAAAAEAABAAAAAAAQAAAAAAAAABAAAIAAAAAAEAACgAAAAAAQAAEAAAAAABAAAAAAAAAAEA\n",
              "AAgAAAAAAQAAKAAAAAABAAAQAAAAAAEAAAAAAAAAAQAACAAAAAABAAAoAAAAAAEAABAAAAAAAQAA\n",
              "AAAAAAABAAAIAAAAAAEAACgAAAAAAQAAEAAAAAABAAAAAAAAAAEAAAgAAAAAAQAAKAAAAAABAAAQ\n",
              "AAAAAAEAAAAAAAAAAQAACAAAAAABAAAgAAAAAAIAAAgAAAAAAQAAKAAAAAABAAAQAAAAAAEAAAAA\n",
              "AAAAAQAACAAAAAABAAAoAAAAAAEAABAAAAAAAQAAAAAAAAABAAAIAAAAAAEAACgAAAAAAQAAEAAA\n",
              "AAABAAAAAAAAAAEAAAgAAAAAAQAAGAAAAAABAAAIAAAAABxzdHNjAAAAAAAAAAEAAAABAAAAZAAA\n",
              "AAEAAAGkc3RzegAAAAAAAAAAAAAAZAAAFRAAAASbAAAD0AAAAysAAATpAAAAswAAAG8AAALKAAAA\n",
              "aQAAAvAAAABYAAADvwAAAH4AAABdAAADnAAAAGgAAABRAAACYQAAAEMAAANbAAAAcAAAAEQAAABB\n",
              "AAADTwAAAH8AAABIAAAAUAAAArsAAAB0AAAAYQAAAD8AAAMcAAAAgAAAAE8AAAB8AAADPgAAAH4A\n",
              "AABQAAAAMwAAAv4AAABtAAAAQQAAAGQAAALdAAAAdAAAAD8AAAA2AAAC/wAAAGIAAABIAAAASAAA\n",
              "Av4AAABrAAAASwAAADEAAAKwAAAAZwAAAD4AAABOAAAC0wAAAGsAAAA9AAAAPAAAAtMAAABiAAAA\n",
              "NQAAAD4AAALCAAAAbwAAADQAAABKAAACUgAAAHkAAABcAAAATgAAAiIAAABRAAAAaAAAAGgAAAH5\n",
              "AAAAUQAAAF8AAABiAAAB1QAAAGUAAABGAAABZgAAAG4AAABbAAAALwAAAWAAAABhAAAARAAAADIA\n",
              "AAENAAAAcAAAADoAAABHAAAAgAAAAEUAAAAUc3RjbwAAAAAAAAABAAAALAAAAGJ1ZHRhAAAAWm1l\n",
              "dGEAAAAAAAAAIWhkbHIAAAAAAAAAAG1kaXJhcHBsAAAAAAAAAAAAAAAALWlsc3QAAAAlqXRvbwAA\n",
              "AB1kYXRhAAAAAQAAAABMYXZmNTcuODMuMTAw\n",
              "\">\n",
              "  Your browser does not support the video tag.\n",
              "</video>"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 274
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ap_60Quy0ELN",
        "colab_type": "text"
      },
      "source": [
        "\n",
        "## Exercise\n",
        "\n",
        "1. MP Neuron: Test and Train size 0.1 (manipulate and see)\n",
        "2. MP Neuron: Binarisation of input using PD cut (we saw by changing [0, 1] to   [1,  0] a better change was seen) treshold value may help in improving the data preparation for training the model\n",
        "3. MP Neuron: Plot try x axis (parameter space) as b and y axis as accuracy. For training data and testing data  (check whether the train and test accuracy are co-related i.e., to check the maximal value for given value of b for test data and train data) \n",
        "4. Perceptron: Try larger values of epoch and learning rate and understand the trade off for getting good accuracy for training the model\n",
        "5. Perceptron: Try random intializer for weights and bias and seed for reproducing the results\n",
        "6. Perceptron: Animate the plot for varying cases of Epochs and learning rate"
      ]
    }
  ]
}