examples/Deep_AutoViML_Text_Test.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Imported deep_auto_viml. version=0.0.53\n",
      "from deep_autoviml import deep_autoviml as deepauto\n",
      "-------------------\n",
      "model, cat_vocab_dict = deepauto.fit(train, target, keras_model_type=\"fast\",\n",
      "\t\tproject_name=\"deep_autoviml\", keras_options=keras_options,  \n",
      "\t\tmodel_options=model_options, save_model_flag=True, use_my_model='',\n",
      "\t\tmodel_use_case='', verbose=0)\n",
      "\n",
      "predictions = deepauto.predict(model, project_name, test_dataset=test,\n",
      "                                 keras_model_type=keras_model_type, \n",
      "                                 cat_vocab_dict=cat_vocab_dict)\n",
      "                                \n"
     ]
    }
   ],
   "source": [
    "from deep_autoviml import deep_autoviml as deepauto"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "project_name=\"deep_autoviml\"\n",
    "#model_options={'text_directory':'aclImdb/'}\n",
    "model_options = {}\n",
    "keras_options={}\n",
    "keras_model_type = \"text\"\n",
    "#text_data_or_file = 'aclImdb/'\n",
    "text_data_or_file = '../Ram/data_sets/NLP_small.csv'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "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>Product_Description</th>\n",
       "      <th>Sentiment</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>The Web DesignerÛªs Guide to iOS (and Android...</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                 Product_Description  Sentiment\n",
       "0  The Web DesignerÛªs Guide to iOS (and Android...          2"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import pandas as pd\n",
    "df = pd.read_csv(text_data_or_file)\n",
    "df = df[[\"Product_Description\",\"Sentiment\"]]\n",
    "df.head(1)"
   ]
  },
  {
   "cell_type": "raw",
   "metadata": {},
   "source": [
    "from deep_autoviml.data_load.extract import load_text_data\n",
    "train_ds, valid_ds, cat_vocab_dict, model_options = load_text_data(text_data_or_file,\n",
    "                                                    project_name=\"deep_autoviml\", \n",
    "                                                    keras_options={}, model_options={'text_directory':'aclImdb/'},\n",
    "                        verbose=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "    Only CPU found on this device\n",
      "Setting CPU strategy using 1 devices\n",
      "    No text directory given. Using train data given in ../Ram/data_sets/NLP_small.csv...\n",
      "CSV file being loaded into tf.data.Dataset\n",
      "    Since number of rows in file <= 10000 maxrows, loading entire file into pandas for EDA\n",
      "    Model type is Multi_Classification and Single_Label problem\n",
      "Alert! Modified column names to satisfy rules for column names in Tensorflow...\n",
      "    Model type is Multi_Classification and Single_Label problem\n",
      "Loaded a small data sample of size = (499, 4) into pandas dataframe to analyze...\n",
      "    Classifying variables using data sample in pandas...\n",
      "############## C L A S S I F Y I N G  V A R I A B L E S  ####################\n",
      "Classifying variables in data set...\n",
      "    3 Predictors classified...\n",
      "    possible latitude columns in dataset: []\n",
      "        after further analysis, no latitude columns found\n",
      "    possible longitude columns in dataset: []\n",
      "        after further analysis, no longitude columns found\n",
      "        1 variable(s) to be removed since they were ID or low-information variables\n",
      "Distribution of string columns in datatset:\n",
      "    number of binary = 0, cats = 0, high cats = 0, very high cats = 1\n",
      "Distribution of integer columns in datatset:\n",
      "    number of binary = 0, cats = 2, high cats = 0, very high cats = 1\n",
      "Distribution of floats: floats = 0\n",
      "Data Transformation Advisory:\n",
      "    transformed ['text_id'] from integer to float\n",
      "    product_description is detected as an NLP variable\n",
      "Not performing feature crossing for categorical nor integer variables\n",
      "Loading your input file(s) data directly into tf.data.Dataset...\n",
      "    train data loaded successfully.\n",
      "Boolean column successfully processed\n",
      "\n",
      "Number of predictors to be used = 2 in predict step: keras preprocessing...\n",
      "Dropping ['text_id'] columns marked for deletion...\n",
      "       Class  -> Counts -> Percent\n",
      "           0:       1  ->    0.2%\n",
      "           1:      35  ->    7.0%\n",
      "           2:     305  ->   61.1%\n",
      "           3:     158  ->   31.7%\n",
      "Loaded text classification file or dataframe using input given:\n",
      "\n",
      "Splitting train into 80+20 percent: train and validation data\n",
      "Tensorboard log directory can be found at: deep_autoviml_text\\mylogs\n",
      "Training text model. This will take time...\n",
      "Epoch 1/20\n",
      "8/8 [==============================] - 2s 111ms/step - loss: 1.6440 - accuracy: 0.1481 - val_loss: 1.6051 - val_accuracy: 0.1826\n",
      "Epoch 2/20\n",
      "8/8 [==============================] - 0s 29ms/step - loss: 1.4847 - accuracy: 0.2675 - val_loss: 1.4741 - val_accuracy: 0.2783\n",
      "Epoch 3/20\n",
      "8/8 [==============================] - 0s 34ms/step - loss: 1.3550 - accuracy: 0.3621 - val_loss: 1.3694 - val_accuracy: 0.3565\n",
      "Epoch 4/20\n",
      "8/8 [==============================] - 0s 29ms/step - loss: 1.2517 - accuracy: 0.4486 - val_loss: 1.2856 - val_accuracy: 0.4609\n",
      "Epoch 5/20\n",
      "8/8 [==============================] - 0s 47ms/step - loss: 1.1697 - accuracy: 0.5473 - val_loss: 1.2189 - val_accuracy: 0.4957\n",
      "Epoch 6/20\n",
      "8/8 [==============================] - 0s 28ms/step - loss: 1.1050 - accuracy: 0.5679 - val_loss: 1.1656 - val_accuracy: 0.5391\n",
      "Epoch 7/20\n",
      "8/8 [==============================] - 0s 32ms/step - loss: 1.0537 - accuracy: 0.5926 - val_loss: 1.1237 - val_accuracy: 0.5652\n",
      "Epoch 8/20\n",
      "8/8 [==============================] - 0s 31ms/step - loss: 1.0131 - accuracy: 0.5967 - val_loss: 1.0907 - val_accuracy: 0.5739\n",
      "Epoch 9/20\n",
      "8/8 [==============================] - 0s 33ms/step - loss: 0.9810 - accuracy: 0.5967 - val_loss: 1.0648 - val_accuracy: 0.5739\n",
      "Epoch 10/20\n",
      "8/8 [==============================] - 0s 29ms/step - loss: 0.9556 - accuracy: 0.6008 - val_loss: 1.0440 - val_accuracy: 0.5826\n",
      "Epoch 11/20\n",
      "8/8 [==============================] - 0s 37ms/step - loss: 0.9349 - accuracy: 0.6132 - val_loss: 1.0266 - val_accuracy: 0.6000\n",
      "Epoch 12/20\n",
      "8/8 [==============================] - 0s 29ms/step - loss: 0.9177 - accuracy: 0.6255 - val_loss: 1.0118 - val_accuracy: 0.6087\n",
      "Epoch 13/20\n",
      "8/8 [==============================] - 0s 35ms/step - loss: 0.9033 - accuracy: 0.6214 - val_loss: 0.9990 - val_accuracy: 0.6000\n",
      "Epoch 14/20\n",
      "8/8 [==============================] - 0s 41ms/step - loss: 0.8908 - accuracy: 0.6173 - val_loss: 0.9880 - val_accuracy: 0.5913\n",
      "Epoch 15/20\n",
      "8/8 [==============================] - 0s 33ms/step - loss: 0.8799 - accuracy: 0.6173 - val_loss: 0.9782 - val_accuracy: 0.5826\n",
      "Epoch 16/20\n",
      "8/8 [==============================] - 0s 29ms/step - loss: 0.8703 - accuracy: 0.6132 - val_loss: 0.9693 - val_accuracy: 0.5826\n",
      "Epoch 17/20\n",
      "8/8 [==============================] - 0s 38ms/step - loss: 0.8618 - accuracy: 0.6214 - val_loss: 0.9613 - val_accuracy: 0.5913\n",
      "Epoch 18/20\n",
      "8/8 [==============================] - 0s 29ms/step - loss: 0.8541 - accuracy: 0.6337 - val_loss: 0.9541 - val_accuracy: 0.6087\n",
      "Epoch 19/20\n",
      "8/8 [==============================] - 0s 30ms/step - loss: 0.8471 - accuracy: 0.6337 - val_loss: 0.9474 - val_accuracy: 0.6087\n",
      "Epoch 20/20\n",
      "8/8 [==============================] - 0s 30ms/step - loss: 0.8409 - accuracy: 0.6379 - val_loss: 0.9413 - val_accuracy: 0.6174\n",
      "4/4 [==============================] - 0s 10ms/step - loss: 0.9413 - accuracy: 0.6174\n",
      "    Model accuracy in text validation data: 0.6173912882804871\n",
      "4/4 [==============================] - 0s 8ms/step - loss: 0.9413 - accuracy: 0.6174\n",
      "Loss:  0.9412603378295898\n",
      "Accuracy:  0.6173912882804871\n",
      "\n",
      "Saving model in deep_autoviml\\text_model now...this will take time...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:absl:Found untraced functions such as _destroy_resource while saving (showing 1 of 1). These functions will not be directly callable after loading.\n",
      "Exception ignored in: <function CapturableResource.__del__ at 0x000001DEB0AE5310>\n",
      "Traceback (most recent call last):\n",
      "  File \"C:\\Users\\Ram\\AppData\\Roaming\\Python\\Python38\\site-packages\\tensorflow\\python\\training\\tracking\\tracking.py\", line 277, in __del__\n",
      "    self._destroy_resource()\n",
      "  File \"C:\\Users\\Ram\\AppData\\Roaming\\Python\\Python38\\site-packages\\tensorflow\\python\\eager\\def_function.py\", line 889, in __call__\n",
      "    result = self._call(*args, **kwds)\n",
      "  File \"C:\\Users\\Ram\\AppData\\Roaming\\Python\\Python38\\site-packages\\tensorflow\\python\\eager\\def_function.py\", line 933, in _call\n",
      "    self._initialize(args, kwds, add_initializers_to=initializers)\n",
      "  File \"C:\\Users\\Ram\\AppData\\Roaming\\Python\\Python38\\site-packages\\tensorflow\\python\\eager\\def_function.py\", line 763, in _initialize\n",
      "    self._stateful_fn._get_concrete_function_internal_garbage_collected(  # pylint: disable=protected-access\n",
      "  File \"C:\\Users\\Ram\\AppData\\Roaming\\Python\\Python38\\site-packages\\tensorflow\\python\\eager\\function.py\", line 3050, in _get_concrete_function_internal_garbage_collected\n",
      "    graph_function, _ = self._maybe_define_function(args, kwargs)\n",
      "  File \"C:\\Users\\Ram\\AppData\\Roaming\\Python\\Python38\\site-packages\\tensorflow\\python\\eager\\function.py\", line 3444, in _maybe_define_function\n",
      "    graph_function = self._create_graph_function(args, kwargs)\n",
      "  File \"C:\\Users\\Ram\\AppData\\Roaming\\Python\\Python38\\site-packages\\tensorflow\\python\\eager\\function.py\", line 3279, in _create_graph_function\n",
      "    func_graph_module.func_graph_from_py_func(\n",
      "  File \"C:\\Users\\Ram\\AppData\\Roaming\\Python\\Python38\\site-packages\\tensorflow\\python\\framework\\func_graph.py\", line 999, in func_graph_from_py_func\n",
      "    func_outputs = python_func(*func_args, **func_kwargs)\n",
      "  File \"C:\\Users\\Ram\\AppData\\Roaming\\Python\\Python38\\site-packages\\tensorflow\\python\\eager\\def_function.py\", line 672, in wrapped_fn\n",
      "    out = weak_wrapped_fn().__wrapped__(*args, **kwds)\n",
      "  File \"C:\\Users\\Ram\\AppData\\Roaming\\Python\\Python38\\site-packages\\tensorflow\\python\\eager\\function.py\", line 3971, in bound_method_wrapper\n",
      "    return wrapped_fn(*args, **kwargs)\n",
      "  File \"C:\\Users\\Ram\\AppData\\Roaming\\Python\\Python38\\site-packages\\tensorflow\\python\\framework\\func_graph.py\", line 986, in wrapper\n",
      "    raise e.ag_error_metadata.to_exception(e)\n",
      "RuntimeError: in user code:\n",
      "\n",
      "    C:\\Users\\Ram\\AppData\\Roaming\\Python\\Python38\\site-packages\\tensorflow\\python\\saved_model\\load.py:655 _destroy_resource  *\n",
      "        raise RuntimeError()\n",
      "\n",
      "    RuntimeError: \n",
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "     deep_autoviml text saved in deep_autoviml\\text_model directory\n",
      "Model: \"sequential_6\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "keras_layer (KerasLayer)     (None, 20)                400020    \n",
      "_________________________________________________________________\n",
      "dense_11 (Dense)             (None, 16)                336       \n",
      "_________________________________________________________________\n",
      "dense_12 (Dense)             (None, 4)                 68        \n",
      "=================================================================\n",
      "Total params: 400,424\n",
      "Trainable params: 404\n",
      "Non-trainable params: 400,020\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "model, cat_vocab_dict = deepauto.fit(text_data_or_file, \"Sentiment\", keras_model_type=keras_model_type,\n",
    "\t\tproject_name=project_name, keras_options=keras_options,  \n",
    "\t\tmodel_options=model_options, save_model_flag=True, use_my_model='',\n",
    "\t\tmodel_use_case='', verbose=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "from deep_autoviml.modeling.predict_model import predict_text"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Using <tensorflow.python.keras.engine.sequential.Sequential object at 0x000001DEC43D04F0> model provided as input...\n",
      "Time taken to load saved model = 0 seconds\n",
      "    loading and predicting on CSV file : ../Ram/data_sets/NLP_small.csv\n",
      "Loaded test data size: 499\n",
      "Alert! Modified column names to satisfy rules for column names in Tensorflow...\n",
      "    test data loaded successfully.\n",
      "    number of steps needed to predict: 16\n",
      "    Sample predictions before inverse_transform: [2 2 2 2 2]\n",
      "    Sample predictions after transform: [2 2 2 2 2]\n"
     ]
    }
   ],
   "source": [
    "test_text_dir = '../Ram/data_sets/NLP_small.csv'\n",
    "predictions = predict_text(test_text_dir, model, cat_vocab_dict)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([2, 2, 2, 2, 2], dtype=int64)"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_probas = predictions[0]\n",
    "y_preds = predictions[-1]\n",
    "y_preds[:5]"
   ]
  },
  {
   "cell_type": "raw",
   "metadata": {},
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import os\n",
    "import re\n",
    "import shutil\n",
    "import string\n",
    "import tensorflow as tf\n",
    "\n",
    "from tensorflow.keras import layers\n",
    "from tensorflow.keras import losses\n",
    "from tensorflow.keras import preprocessing\n",
    "from tensorflow.keras.layers.experimental.preprocessing import TextVectorization\n",
    "print(tf.__version__)"
   ]
  },
  {
   "cell_type": "raw",
   "metadata": {},
   "source": [
    "url = \"https://ai.stanford.edu/~amaas/data/sentiment/aclImdb_v1.tar.gz\"\n",
    "\n",
    "dataset = tf.keras.utils.get_file(\"aclImdb_v1\", url,\n",
    "                                    untar=True, cache_dir='.',\n",
    "                                    cache_subdir='')\n",
    "\n",
    "dataset_dir = os.path.join(os.path.dirname(dataset), 'aclImdb')"
   ]
  },
  {
   "cell_type": "raw",
   "metadata": {},
   "source": [
    "os.listdir(dataset_dir)"
   ]
  },
  {
   "cell_type": "raw",
   "metadata": {},
   "source": [
    "train_dir = os.path.join(dataset_dir, 'train')\n",
    "os.listdir(train_dir)"
   ]
  },
  {
   "cell_type": "raw",
   "metadata": {},
   "source": [
    "sample_file = os.path.join(train_dir, 'pos/1181_9.txt')\n",
    "with open(sample_file) as f:\n",
    "  print(f.read())"
   ]
  },
  {
   "cell_type": "raw",
   "metadata": {},
   "source": [
    "remove_dir = os.path.join(train_dir, 'unsup')\n",
    "shutil.rmtree(remove_dir)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}