jupyter_russian/topic10_boosting/lesson10_part2_xgboost_scikit_gboost.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<center>\n",
    "<img src=\"../../img/ods_stickers.jpg\">\n",
    "## Открытый курс по машинному обучению. Сессия № 2\n",
    "Автор материала: программист-исследователь Mail.ru Group, старший преподаватель Факультета Компьютерных Наук ВШЭ Юрий Кашницкий. Материал распространяется на условиях лицензии [Creative Commons CC BY-NC-SA 4.0](https://creativecommons.org/licenses/by-nc-sa/4.0/). Можно использовать в любых целях (редактировать, поправлять и брать за основу), кроме коммерческих, но с обязательным упоминанием автора материала."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# <center>Тема 10. Бустинг\n",
    "## <center>Часть 2. Сравнение Xgboost и градиентного бустинга Sklearn</center>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "## XGBoost\n",
    "\n",
    "Выделяют три группы параметров:\n",
    "- Общие параметры, отвечающие за базовый алгоритм для бустинга и распараллеливание.\n",
    "- Параметры выбранного базового алгоритма.\n",
    "- Параметры обучения, отвечающие за функцию потерь и метрику качества на валидации.\n",
    "\n",
    "**1. Общие параметры:**\n",
    "- booster [default=gbtree] - тип базового алгоритма для бустинга: дерево решений gbtree или линейная модель gblinear. \n",
    "- silent [default=0] - выдавать (silent=0) или нет (silent=1) сообщения по ходу работы алгоритма.\n",
    "- nthread [default to maximum number of threads available if not set] - число нитей доступных для параллельной работы xgboost.\n",
    "\n",
    "**2. Параметры базового алгоритма:**\n",
    "\n",
    "**2.1. Дерево решений:**\n",
    "- eta [default=0.3] - темп обучения, перед добавлением дерева в композицию оно умножается на eta. Используется для предотвращения переобучения за счёт \"сокращения\" весов базовых алгоритмов, делая модель более консервативной. Чем меньше eta, тем больше нужно итераций num_boost_round для обучения модели с хорошим качеством. Диапазон: [0, 1]\n",
    "- gamma [default=0] - минимальное снижение значения функции потерь, необходимое для дальнейшего разбиения вершины дерева. Большие значения gamma > 0 приводят к более консервативным моделям. Диапазон: [0, $\\infty$).\n",
    "- max_depth [default=6] - максимальная глубина дерева. Диапазон: [1, $\\infty$). \n",
    "- min_child_weight [default=1] - минимальное необходимое (взвешенное) число примеров в каждой вершине. Чем больше, тем более консервативна итоговая модель. Диапазон: [0, $\\infty$).\n",
    "- max_delta_step [default=0] - обычно равен нулю. Положительные значения используются при несбалансированных классах для ускорения сходимости. Диапазон [0, $\\infty$).\n",
    "- subsample [default=1] - доля выборки, используемая для обучения каждого дерева. Если subsample < 1, то выбирается случайная подвыборка, что помогает в борьбе с переобучением. Диапазон: (0, 1]\n",
    "- colsample_bytree [default=1] - доля признаков, используемая для обучения каждого дерева. Диапазон: (0, 1]\n",
    "- lambda [default=1] - коэффициент перед $L_2$-регуляризатором в функции потерь.\n",
    "- alpha [default=0] - коэффициент перед $L_1$-регуляризатором в функции потерь.\n",
    "\n",
    "**2.2. Линейная модель:**\n",
    "- lambda [default=0] - коэффициент перед $L_2$-регуляризатором вектора весов в функции потерь.\n",
    "- alpha [default=0] - коэффициент перед $L_1$-регуляризатором вектора весов в функции потерь.\n",
    "- lambda_bias [default=0] - коэффициент перед $L_2$-регуляризатором смещения (свободного члена) в функции потерь.\n",
    "\n",
    "**3. Параметры задачи обучения:**\n",
    "- objective [default=reg:linear] - используемая при обучении функция потерь:\n",
    "    - \"reg:linear\" – линейная регрессия.\n",
    "    - \"reg:logistic\" – логистическая регрессия.\n",
    "    - \"binary:logistic\" – логистическая регрессия для бинарной классификации, на выходе - вероятность.\n",
    "    - \"binary:logitraw\" – то же самое, но на выходе - значение до его преобразования логистической функцией.\n",
    "    - \"count:poisson\" – регрессия Пуассона (используется для оценки числа каких-то событий, счётный признак), на выходе - матожидания распределения Пуассона. В этом случае max_delta_step автоматически устанавливается равным 0.7.\n",
    "    - \"multi:softmax\" – обобщение логистической регрессии на многоклассовый случай. При этом нужно задать параметр num_class.\n",
    "    - \"multi:softprob\" – то же самое, но на выходе - вектор размера ndata * nclass, который можно преобразовать в матрицу, содержащую вероятности отнесения данного объекта к данному классу.\n",
    "    - \"rank:pairwise\" – используется для задач ранжирования.\n",
    "- base_score [default=0.5] - инициализация значения модели для всех примеров, глобальное смещение.\n",
    "- eval_metric [default according to objective] - метрика качества на валидационной выборке (по умолчанию соответствует функции потерь: rmse - для регрессии, error - для классификации, mean average precision - для ранжирования). Выбрать можно одну из следующих метрик:\n",
    "    - \"rmse\": root mean square error.\n",
    "    - \"logloss\": минус логарифм правдоподобия.\n",
    "    - \"error\": доля ошибок для бинарной классификации.\n",
    "    - \"merror\": то же самое для многоклассовой классификации.\n",
    "    - \"mlogloss\": logloss для многоклассовой классификации.\n",
    "    - \"auc\": AUC.\n",
    "    - \"ndcg\": Normalized Discounted Cumulative Gain.\n",
    "    - \"map\": Mean average precision.\n",
    "    - \"ndcg@n\",”map@n”: здесь n - целое число, первые n позиций в списке не учитываются.\n",
    "    - \"ndcg-\",”map-”,”ndcg@n-”,”map@n-”: списку из всех положительных примеров будет присвоено значение 0 (вместо 1).\n",
    "- seed [default=0] - для воспроизводимости \"случайности\".\n",
    "\n",
    "**Параметры в xgboost.train**:\n",
    "- params (dict) – параметры, описанные выше.\n",
    "- dtrain (DMatrix) – обучающая выборка.\n",
    "- num_boost_round (int) – число итераций бустинга.\n",
    "- evals (list) – список для оценки качества во время обучения.\n",
    "- obj (function) – собственная функция потерь.\n",
    "- feval (function) – собственная функция для оценки качества.\n",
    "- maximize (bool) – нужно ли максимизировать feval.\n",
    "- early_stopping_rounds (int) – активирует early stopping. Ошибка на валидации должна уменьшаться каждые early_stopping_rounds итераций для продолжения обучения. Список evals должен быть не пуст. Возвращается модель с последней итерации. Если произошел ранний останов, то модель будет содержать поля: bst.best_score и bst.best_iteration.\n",
    "- evals_result (dict) – результаты оценки качества.\n",
    "- verbose_eval (bool) – вывод значения метрики качества на каждой итерации бустинга.\n",
    "- learning_rates (list or function) – коэффициент скорости обучения для каждой итерации - list l: eta = l[boosting round] - function f: eta = f(boosting round, num_boost_round).\n",
    "- xgb_model (file name of stored xgb model or ‘Booster’ instance) – возможность продолжить обучения имеющейся модели XGB.\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## sklearn.ensemble.GradientBoostingClassifier\n",
    "- loss [default=\"deviance\"] - оптимизируемая функция потерь.  Одна из {\"deviance\", \"exponential\"}. Первая соответствует логистической регрессии и возвращает вероятности, вторая - AdaBoost.\n",
    "- learning_rate [default=0.1] - темп обучения, аналогично eta для XGBoost.\n",
    "- n_estimators [default=100] - число итераций градиентного бустинга.\n",
    "- max_depth [default=3] - аналогично max_depth для XGBoost.\n",
    "- min_samples_split [default=2] - минимальное число примеров, необходимое для разветвления в данной вершине,  аналогично min_child_weight для XGBoost.\n",
    "- min_samples_leaf [default=1] - минимальное число примеров в листе.\n",
    "- min_weight_fraction_leaf [default=0.0] - минимальное взвешенное число примеров в листе.\n",
    "- subsample [default=1.0] - аналогично subsample для XGBoost.\n",
    "- max_features (int, float, string or None) [default=None] - число (или доля) признаков, используемых при разбиении вершины.\n",
    "    - \"auto\", тогда max_features=sqrt(n_features).\n",
    "    - \"sqrt\", тогда max_features=sqrt(n_features).\n",
    "    - \"log2\", тогда max_features=log2(n_features).\n",
    "    - None, тогда max_features=n_features.\n",
    "- max_leaf_nodes [default=None]\n",
    "- init (BaseEstimator or None) [default=None] - алгоритм для начальных предсказаний.\n",
    "- verbose [default=0] - аналогично silent для XGBoost.\n",
    "- warm_start [default=False] - если True, используется ансамбль с предыдущего вызова fit, новые алгоритмы добавляются к нему, иначе строится новый алгоритм."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Сравнение алгоритмов по времени работы\n",
    "\n",
    "Посмотрим на время обучения классификаторов XGBooster и GradientBoostingClassifier. Для этого будем генерировать выборку из 1000 объектов и 50 признаков с помощью sklearn.datasets.make_classification и замерять время обучения."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "dae55ae373c14723b04d01d135ac5ae3"
      }
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "import time\n",
    "from tqdm import tqdm_notebook\n",
    "import xgboost as xgb\n",
    "from sklearn.ensemble import GradientBoostingClassifier\n",
    "from sklearn.datasets import make_classification\n",
    "\n",
    "xgb_params1 = {'booster': 'gbtree', 'max_depth': 3, 'eta': 0.1, \n",
    "    'silent': 1, 'objective': 'binary:logistic', 'nthread': 1}\n",
    "xgb_params2 = {'booster': 'gbtree', 'max_depth': 3, 'eta': 0.1, \n",
    "    'silent': 1, 'objective': 'binary:logistic', 'nthread': 4}\n",
    "sklearn_params = {'n_estimators': 100, 'max_depth': 3}\n",
    "\n",
    "xgb_time1 = list()\n",
    "xgb_time2 = list()\n",
    "sklearn_time = list()\n",
    "\n",
    "n_runs = 50\n",
    "\n",
    "for i in tqdm_notebook(range(n_runs)):\n",
    "    # Generating dataset\n",
    "    X, y = make_classification(n_samples=1000, n_features=50, \n",
    "                               n_informative=20)\n",
    "    # Training XGBooster (nthread=1)\n",
    "    t = time.time()\n",
    "    bst = xgb.train(xgb_params1, xgb.DMatrix(X, label=y), \n",
    "                    num_boost_round=100)\n",
    "    elapsed = time.time() - t\n",
    "    xgb_time1.append(elapsed)\n",
    "    # Training XGBooster (nthread=4)\n",
    "    t = time.time()\n",
    "    bst = xgb.train(xgb_params2, xgb.DMatrix(X, label=y), \n",
    "                    num_boost_round=100)\n",
    "    elapsed = time.time() - t\n",
    "    xgb_time2.append(elapsed)\n",
    "    # Training GradientBoostingClassifier\n",
    "    t = time.time()\n",
    "    clf = GradientBoostingClassifier(**sklearn_params).fit(X, y)\n",
    "    elapsed = time.time() - t\n",
    "    sklearn_time.append(elapsed)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAtoAAAHwCAYAAACYMcj+AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nJWGmOgAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1149c8c18>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "%matplotlib  inline\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "xgb_mean1 = sum(xgb_time1) / n_runs\n",
    "xgb_mean2 = sum(xgb_time2) / n_runs\n",
    "sklearn_mean = sum(sklearn_time) / n_runs\n",
    "\n",
    "plt.figure(figsize=(12, 8))\n",
    "plt.plot(xgb_time1, label='XGBooster (nthread = 1)', lw=2)\n",
    "plt.plot(xgb_time2, label='XGBooster (nthread = 4)', lw=2)\n",
    "plt.plot(sklearn_time, label='GradientBoostingClassifier', lw=2)\n",
    "plt.legend(loc='best')\n",
    "plt.text(1, (xgb_mean1 + sklearn_mean) / 2, \n",
    "         'XGBoost (nthread = 1) mean time = %.2f' % xgb_mean1 +\n",
    "         '\\n\\nXGBoost (nthread = 4) mean time = %.2f' % xgb_mean2 +\n",
    "         '\\n\\nScikit-learn mean time = %.2f' % sklearn_mean,\n",
    "            fontsize = 15)\n",
    "plt.xlabel('Iteration number')\n",
    "plt.ylabel('Running time, sec')\n",
    "plt.title('XGBoost vs. GradientBoostingClassifier comparison')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Выводы\n",
    "\n",
    "Основные преимущества XGBoost по сравнению с sklearn.ensembles.GradientBoostingClassifier:\n",
    "- Помимо деревьев возможно использование линейных моделей в качестве базовых классификаторов.\n",
    "- Скорость работы.\n",
    "- Возможность распараллеливания.\n",
    "- Значительно больший выбор стандартных функций потерь, а также возможность задавать свою функцию потерь.\n",
    "- Наличие регуляризаторов в итоговой функции потерь и возможность задавать их коэффициенты, что даёт еще один метод борьбы с переобучением, помимо использования случайности (subsample, colsample_bytree) и основных параметров дерева решений.\n",
    "- Встроенная обработка missing values."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Полезные ссылки\n",
    "- [XGBoost](https://xgboost.readthedocs.org/en/latest/parameter.html)\n",
    "- [GradientBoostingClassifier](http://scikit-learn.org/stable/modules/generated/sklearn.ensemble.GradientBoostingClassifier.html)\n",
    "- [Сравнение](https://github.com/szilard/benchm-ml) различных библиотек для машинного обучения, в том числе sklearn и xgboost "
   ]
  }
 ],
 "metadata": {
  "anaconda-cloud": {},
  "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": "3.6.1"
  },
  "name": "2_1_7_regul.ipynb"
 },
 "nbformat": 4,
 "nbformat_minor": 1
}