0.机器学习集成模型.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "d1c110851ac8fd1b",
   "metadata": {},
   "outputs": [],
   "source": [
    "# 构建机器学习模型并训练：回归模型、分类模型、模型中特征重要性可视化、调参\n",
    "# python3.9 + sklearn + numpy + scipy + pandas + matplotlib + hyperopt (调参数) + bayesian-optimization (调参数 from bayes_opt import BayesianOptimization) + catboost （需要gpu）+ tensorflow 2.13 (keras)\n",
    "# lightgbm + xgboost\n",
    "# pip install bayesian-optimization=1.4.0\n",
    "\n",
    "# 回归模型：LinearRegression、Ridge、Lasso、ElasticNet、SVR、KNeighborsRegressor、DecisionTreeRegressor、RandomForestRegressor、GradientBoostingRegressor、xgboost.XGBRegressor、lightgbm.LGBMRegressor。\n",
    "# 分类模型：LogisticRegression、KNeighborsClassifier、GaussianNB、tree、BaggingClassifier、RandomForestClassifier、ExtraTreesClassifier、AdaBoostClassifier、GradientBoostingClassifier、lightgbm.LGBMClassifier、xgboost.XGBClassifier。\n",
    "# 模型习得特征重要性的可视化。\n",
    "# 正则化：L1、L2、elasticnet。\n",
    "# 调参：GridSearchCV、RandomizedSearchCV。\n",
    "\n",
    "# 重要模型：lightgbm模型、catboost、xgboost、RF。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "36226c3d",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-08-25T01:28:31.375198700Z",
     "start_time": "2023-08-25T01:28:30.331723100Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "acc = 0.926 auc = 0.49571734475374735 mcc = -0.023871938507169333\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0       0.93      0.99      0.96       467\n",
      "         1.0       0.00      0.00      0.00        33\n",
      "\n",
      "    accuracy                           0.93       500\n",
      "   macro avg       0.47      0.50      0.48       500\n",
      "weighted avg       0.87      0.93      0.90       500\n",
      "\n",
      "Confusion matrix, without normalization\n",
      "[[463   4]\n",
      " [ 33   0]]\n",
      "Normalized confusion matrix\n",
      "[[0.99 0.01]\n",
      " [1.   0.  ]]\n"
     ]
    },
    {
     "data": {
      "text/plain": "<Figure size 640x480 with 2 Axes>",
      "image/png": "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\n"
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": "<Figure size 640x480 with 2 Axes>",
      "image/png": "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\n"
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# 分类评价：混淆矩阵、auc等\n",
    "# 数据集2：分类\n",
    "\n",
    "import itertools\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.metrics import confusion_matrix, classification_report\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.metrics import roc_auc_score, accuracy_score, matthews_corrcoef\n",
    "import warnings\n",
    "warnings.filterwarnings(\"ignore\")\n",
    "%matplotlib inline\n",
    "\n",
    "# label name\n",
    "class_names = ['no-repeat', 'repeat']\n",
    "\n",
    "# Split the data into a training set and a test set\n",
    "X_train, X_test, y_train, y_test = train_test_split(train, target, random_state=0)\n",
    "\n",
    "# Run classifier, using a model that is too regularized (C too low) to see\n",
    "# the impact on the results\n",
    "clf = RandomForestClassifier(n_jobs=-1)\n",
    "y_pred = clf.fit(X_train, y_train).predict(X_test)\n",
    "# 不同评价标准下的结果\n",
    "acc = accuracy_score(y_test, y_pred)\n",
    "auc = roc_auc_score(y_test, y_pred)\n",
    "mcc = matthews_corrcoef(y_test, y_pred)\n",
    "print('acc =', acc, 'auc =', auc, 'mcc =', mcc)\n",
    "print(classification_report(y_test, y_pred))\n",
    "\n",
    "\n",
    "def plot_confusion_matrix(cm, classes, normalize=False, title='Confusion matrix', cmap=plt.cm.Blues):\n",
    "    \"\"\"\n",
    "    This function prints and plots the confusion matrix.\n",
    "    Normalization can be applied by setting `normalize=True`.\n",
    "    \"\"\"\n",
    "    if normalize:\n",
    "        cm = cm.astype('float') / cm.sum(axis=1)[:, np.newaxis]\n",
    "        print(\"Normalized confusion matrix\")\n",
    "    else:\n",
    "        print('Confusion matrix, without normalization')\n",
    "\n",
    "    print(cm)\n",
    "\n",
    "    plt.imshow(cm, interpolation='nearest', cmap=cmap)\n",
    "    plt.title(title)\n",
    "    plt.colorbar()\n",
    "    tick_marks = np.arange(len(classes))\n",
    "    plt.xticks(tick_marks, classes, rotation=45)\n",
    "    plt.yticks(tick_marks, classes)\n",
    "\n",
    "    fmt = '.2f' if normalize else 'd'\n",
    "    thresh = cm.max() / 2.\n",
    "    for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):\n",
    "        plt.text(j, i, format(cm[i, j], fmt), horizontalalignment=\"center\", color=\"white\" if cm[i, j] > thresh else \"black\")\n",
    "\n",
    "    plt.ylabel('True label')\n",
    "    plt.xlabel('Predicted label')\n",
    "    plt.tight_layout()\n",
    "\n",
    "\n",
    "# Compute confusion matrix\n",
    "cnf_matrix = confusion_matrix(y_test, y_pred)\n",
    "np.set_printoptions(precision=2)\n",
    "\n",
    "# Plot non-normalized confusion matrix\n",
    "plt.figure()\n",
    "plot_confusion_matrix(cnf_matrix, classes=class_names, title='Confusion matrix, without normalization')\n",
    "\n",
    "# Plot normalized confusion matrix\n",
    "plt.figure()\n",
    "plot_confusion_matrix(cnf_matrix, classes=class_names, normalize=True, title='Normalized confusion matrix')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "87e572b125302f93",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-08-24T05:34:07.968209900Z",
     "start_time": "2023-08-24T05:34:07.905712500Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Wall time: 0 ns\n",
      "LinearRegression:    0.2642337917628173\n"
     ]
    }
   ],
   "source": [
    "# 线性回归模型 LinearRegression\n",
    "# 数据1：回归\n",
    "\n",
    "from sklearn.linear_model import LinearRegression\n",
    "from sklearn.metrics import mean_squared_error\n",
    "\n",
    "clf = LinearRegression()\n",
    "%time clf.fit(train_data, train_target)\n",
    "test_pred = clf.predict(test_data)\n",
    "score = mean_squared_error(test_target, clf.predict(test_data))\n",
    "print(\"LinearRegression:   \", score)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "id": "d183298605fabb45",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-08-24T07:22:37.836579600Z",
     "start_time": "2023-08-24T07:22:37.758460Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Wall time: 0 ns\n",
      "Ridge:    0.26390267274268325\n"
     ]
    }
   ],
   "source": [
    "# 岭回归模型 Ridge\n",
    "# 数据1：回归\n",
    "\n",
    "from sklearn.linear_model import Ridge\n",
    "from sklearn.metrics import mean_squared_error\n",
    "\n",
    "clf = Ridge()\n",
    "%time clf.fit(train_data, train_target)\n",
    "test_pred = clf.predict(test_data)\n",
    "score = mean_squared_error(test_target, clf.predict(test_data))\n",
    "print(\"Ridge:   \", score)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "id": "2844694c86aee9ef",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-08-24T07:23:12.836557200Z",
     "start_time": "2023-08-24T07:23:12.741743700Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Wall time: 76.2 ms\n",
      "Lasso:    1.0272506955811558\n"
     ]
    }
   ],
   "source": [
    "# Lasso回归模型 Lasso\n",
    "# 数据1：回归\n",
    "\n",
    "from sklearn.linear_model import Lasso\n",
    "from sklearn.metrics import mean_squared_error\n",
    "\n",
    "clf = Lasso()\n",
    "%time clf.fit(train_data, train_target)\n",
    "test_pred = clf.predict(test_data)\n",
    "score = mean_squared_error(test_target, clf.predict(test_data))\n",
    "print(\"Lasso:   \", score)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "id": "58cfcb5d0d078ecb",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-08-24T07:23:45.445339600Z",
     "start_time": "2023-08-24T07:23:45.378770200Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Wall time: 75.3 µs\n",
      "ElasticNet:    1.0272506955811558\n"
     ]
    }
   ],
   "source": [
    "# ElasticNet回归模型 ElasticNet\n",
    "# 数据1：回归\n",
    "\n",
    "from sklearn.linear_model import ElasticNet\n",
    "from sklearn.metrics import mean_squared_error\n",
    "\n",
    "clf = ElasticNet()\n",
    "%time clf.fit(train_data, train_target)\n",
    "test_pred = clf.predict(test_data)\n",
    "score = mean_squared_error(test_target, clf.predict(test_data))\n",
    "print(\"ElasticNet:   \", score)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "id": "2243b9888b9101fe",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-08-24T07:25:02.082553900Z",
     "start_time": "2023-08-24T07:25:00.671073Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Wall time: 430 ms\n",
      "SVR:    0.216637323528845\n"
     ]
    }
   ],
   "source": [
    "# SVR回归模型 SVR\n",
    "# 数据1：回归\n",
    "\n",
    "from sklearn.svm import SVR\n",
    "from sklearn.metrics import mean_squared_error\n",
    "\n",
    "clf = SVR()\n",
    "%time clf.fit(train_data, train_target)\n",
    "test_pred = clf.predict(test_data)\n",
    "score = mean_squared_error(test_target, clf.predict(test_data))\n",
    "print(\"SVR:   \", score)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "ed74b01206bb2abf",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-08-24T05:34:08.061951200Z",
     "start_time": "2023-08-24T05:34:07.952586600Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Wall time: 0 ns\n",
      "KNeighborsRegressor:    0.27619208861976163\n"
     ]
    }
   ],
   "source": [
    "# K近邻回归模型算法\n",
    "# 数据1：回归\n",
    "\n",
    "from sklearn.neighbors import KNeighborsRegressor\n",
    "from sklearn.metrics import mean_squared_error\n",
    "\n",
    "clf = KNeighborsRegressor(n_neighbors=3)  # 最近三个\n",
    "%time clf.fit(train_data, train_target)\n",
    "test_pred = clf.predict(test_data)\n",
    "score = mean_squared_error(test_target, clf.predict(test_data))\n",
    "print(\"KNeighborsRegressor:   \", score)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "159b9efb50fe826",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-08-24T03:38:24.888326700Z",
     "start_time": "2023-08-24T03:38:24.810176700Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Wall time: 62.5 ms\n",
      "DecisionTreeRegressor:    0.6050157525951557\n"
     ]
    }
   ],
   "source": [
    "# 决策回归树模型算法\n",
    "# 数据1：回归\n",
    "\n",
    "from sklearn.tree import DecisionTreeRegressor\n",
    "from sklearn.metrics import mean_squared_error\n",
    "\n",
    "clf = DecisionTreeRegressor()\n",
    "%time clf.fit(train_data, train_target)\n",
    "test_pred = clf.predict(test_data)\n",
    "score = mean_squared_error(test_target, clf.predict(test_data))\n",
    "print(\"DecisionTreeRegressor:   \", score)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "7608e8688a4e21e3",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-08-24T03:38:35.550659900Z",
     "start_time": "2023-08-24T03:38:30.092201600Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Wall time: 5.32 s\n",
      "RandomForestRegressor:    0.24614504410601207\n"
     ]
    }
   ],
   "source": [
    "# 随机森林回归树模型算法 (**)\n",
    "# 数据1：回归\n",
    "\n",
    "from sklearn.ensemble import RandomForestRegressor\n",
    "from sklearn.metrics import mean_squared_error\n",
    "\n",
    "clf = RandomForestRegressor(n_estimators=200) # 200棵树模型\n",
    "%time clf.fit(train_data, train_target)\n",
    "test_pred = clf.predict(test_data)\n",
    "score = mean_squared_error(test_target, clf.predict(test_data))\n",
    "print(\"RandomForestRegressor:   \", score)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "id": "85e1dfa678fd451c",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-08-24T07:27:13.534662Z",
     "start_time": "2023-08-24T07:27:12.319800300Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Wall time: 1.16 s\n",
      "GradientBoostingRegressor:    0.27728686198310526\n"
     ]
    }
   ],
   "source": [
    "# GBDT模型 GradientBoostingRegressor (*)\n",
    "# 数据1：回归\n",
    "\n",
    "from sklearn.ensemble import GradientBoostingRegressor\n",
    "from sklearn.metrics import mean_squared_error\n",
    "\n",
    "clf = GradientBoostingRegressor()\n",
    "%time clf.fit(train_data, train_target)\n",
    "test_pred = clf.predict(test_data)\n",
    "score = mean_squared_error(test_target, clf.predict(test_data))\n",
    "print(\"GradientBoostingRegressor:   \", score)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "924020dce85581ab",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 2.35 s, sys: 631 ms, total: 2.98 s\n",
      "Wall time: 472 ms\n",
      "XGBRegressor:    0.26703937432297636\n"
     ]
    }
   ],
   "source": [
    "# XGB模型 XGBRegressor (**)\n",
    "# 数据1：回归\n",
    "\n",
    "from xgboost import XGBRegressor\n",
    "from sklearn.metrics import mean_squared_error\n",
    "\n",
    "clf = XGBRegressor(objective='reg:squarederror')\n",
    "%time clf.fit(train_data, train_target)\n",
    "test_pred = clf.predict(test_data)\n",
    "score = mean_squared_error(test_target, clf.predict(test_data))\n",
    "print(\"XGBRegressor:   \", score)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "64299f69f58e06ff",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-08-24T03:40:33.819165Z",
     "start_time": "2023-08-24T03:40:30.566932400Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 34.3 s, sys: 2.24 s, total: 36.5 s\n",
      "Wall time: 4.73 s\n",
      "lightGbm:    0.23134714151049274\n"
     ]
    }
   ],
   "source": [
    "# lgb回归模型 (***)\n",
    "# 数据1：回归\n",
    "# 如果pip3 install lightgbm安装有问题，则lightgbm在conda里安装\n",
    "\n",
    "from lightgbm import LGBMRegressor #lightGbm模型\n",
    "from sklearn.metrics import mean_squared_error\n",
    "clf = LGBMRegressor(\n",
    "    learning_rate=0.01,\n",
    "    max_depth=-1,\n",
    "    n_estimators=5000,\n",
    "    boosting_type='gbdt',\n",
    "    random_state=2019,\n",
    "    objective='regression',\n",
    ")\n",
    "# 训练模型\n",
    "%time clf.fit(X=train_data, y=train_target, eval_metric='MSE', verbose=50)\n",
    "score = mean_squared_error(test_target, clf.predict(test_data))\n",
    "print(\"lightGbm:   \", score)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "32071dff",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 2.15 s, sys: 39.1 ms, total: 2.19 s\n",
      "Wall time: 327 ms\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.926"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# LR模型 LogisticRegression\n",
    "# 数据2：分类\n",
    "\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "\n",
    "stdScaler = StandardScaler()\n",
    "X = stdScaler.fit_transform(train)\n",
    "# Split the data into a training set and a test set\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, target, random_state=0)\n",
    "clf = LogisticRegression(random_state=0, solver='lbfgs', multi_class='multinomial')\n",
    "%time clf.fit(X_train, y_train)\n",
    "clf.score(X_test, y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "8b821b67",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 981 µs, sys: 56 µs, total: 1.04 ms\n",
      "Wall time: 1.05 ms\n"
     ]
    },
    {
     "ename": "AttributeError",
     "evalue": "dlsym(0x9ade32b0, openblas_get_parallel): symbol not found",
     "output_type": "error",
     "traceback": [
      "\u001B[0;31m---------------------------------------------------------------------------\u001B[0m",
      "\u001B[0;31mAttributeError\u001B[0m                            Traceback (most recent call last)",
      "Input \u001B[0;32mIn [3]\u001B[0m, in \u001B[0;36m<cell line: 14>\u001B[0;34m()\u001B[0m\n\u001B[1;32m     12\u001B[0m clf \u001B[38;5;241m=\u001B[39m KNeighborsClassifier(n_neighbors\u001B[38;5;241m=\u001B[39m\u001B[38;5;241m3\u001B[39m)\n\u001B[1;32m     13\u001B[0m get_ipython()\u001B[38;5;241m.\u001B[39mrun_line_magic(\u001B[38;5;124m'\u001B[39m\u001B[38;5;124mtime\u001B[39m\u001B[38;5;124m'\u001B[39m, \u001B[38;5;124m'\u001B[39m\u001B[38;5;124mclf.fit(X_train, y_train)\u001B[39m\u001B[38;5;124m'\u001B[39m)\n\u001B[0;32m---> 14\u001B[0m \u001B[43mclf\u001B[49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43mscore\u001B[49m\u001B[43m(\u001B[49m\u001B[43mX_test\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[43my_test\u001B[49m\u001B[43m)\u001B[49m\n",
      "File \u001B[0;32m~/opt/anaconda3/lib/python3.9/site-packages/sklearn/base.py:666\u001B[0m, in \u001B[0;36mClassifierMixin.score\u001B[0;34m(self, X, y, sample_weight)\u001B[0m\n\u001B[1;32m    641\u001B[0m \u001B[38;5;124;03m\"\"\"\u001B[39;00m\n\u001B[1;32m    642\u001B[0m \u001B[38;5;124;03mReturn the mean accuracy on the given test data and labels.\u001B[39;00m\n\u001B[1;32m    643\u001B[0m \n\u001B[0;32m   (...)\u001B[0m\n\u001B[1;32m    662\u001B[0m \u001B[38;5;124;03m    Mean accuracy of ``self.predict(X)`` wrt. `y`.\u001B[39;00m\n\u001B[1;32m    663\u001B[0m \u001B[38;5;124;03m\"\"\"\u001B[39;00m\n\u001B[1;32m    664\u001B[0m \u001B[38;5;28;01mfrom\u001B[39;00m \u001B[38;5;21;01m.\u001B[39;00m\u001B[38;5;21;01mmetrics\u001B[39;00m \u001B[38;5;28;01mimport\u001B[39;00m accuracy_score\n\u001B[0;32m--> 666\u001B[0m \u001B[38;5;28;01mreturn\u001B[39;00m accuracy_score(y, \u001B[38;5;28;43mself\u001B[39;49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43mpredict\u001B[49m\u001B[43m(\u001B[49m\u001B[43mX\u001B[49m\u001B[43m)\u001B[49m, sample_weight\u001B[38;5;241m=\u001B[39msample_weight)\n",
      "File \u001B[0;32m~/opt/anaconda3/lib/python3.9/site-packages/sklearn/neighbors/_classification.py:219\u001B[0m, in \u001B[0;36mKNeighborsClassifier.predict\u001B[0;34m(self, X)\u001B[0m\n\u001B[1;32m    203\u001B[0m \u001B[38;5;124;03m\"\"\"Predict the class labels for the provided data.\u001B[39;00m\n\u001B[1;32m    204\u001B[0m \n\u001B[1;32m    205\u001B[0m \u001B[38;5;124;03mParameters\u001B[39;00m\n\u001B[0;32m   (...)\u001B[0m\n\u001B[1;32m    214\u001B[0m \u001B[38;5;124;03m    Class labels for each data sample.\u001B[39;00m\n\u001B[1;32m    215\u001B[0m \u001B[38;5;124;03m\"\"\"\u001B[39;00m\n\u001B[1;32m    216\u001B[0m \u001B[38;5;28;01mif\u001B[39;00m \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39mweights \u001B[38;5;241m==\u001B[39m \u001B[38;5;124m\"\u001B[39m\u001B[38;5;124muniform\u001B[39m\u001B[38;5;124m\"\u001B[39m:\n\u001B[1;32m    217\u001B[0m     \u001B[38;5;66;03m# In that case, we do not need the distances to perform\u001B[39;00m\n\u001B[1;32m    218\u001B[0m     \u001B[38;5;66;03m# the weighting so we do not compute them.\u001B[39;00m\n\u001B[0;32m--> 219\u001B[0m     neigh_ind \u001B[38;5;241m=\u001B[39m \u001B[38;5;28;43mself\u001B[39;49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43mkneighbors\u001B[49m\u001B[43m(\u001B[49m\u001B[43mX\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[43mreturn_distance\u001B[49m\u001B[38;5;241;43m=\u001B[39;49m\u001B[38;5;28;43;01mFalse\u001B[39;49;00m\u001B[43m)\u001B[49m\n\u001B[1;32m    220\u001B[0m     neigh_dist \u001B[38;5;241m=\u001B[39m \u001B[38;5;28;01mNone\u001B[39;00m\n\u001B[1;32m    221\u001B[0m \u001B[38;5;28;01melse\u001B[39;00m:\n",
      "File \u001B[0;32m~/opt/anaconda3/lib/python3.9/site-packages/sklearn/neighbors/_base.py:763\u001B[0m, in \u001B[0;36mKNeighborsMixin.kneighbors\u001B[0;34m(self, X, n_neighbors, return_distance)\u001B[0m\n\u001B[1;32m    756\u001B[0m use_pairwise_distances_reductions \u001B[38;5;241m=\u001B[39m (\n\u001B[1;32m    757\u001B[0m     \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39m_fit_method \u001B[38;5;241m==\u001B[39m \u001B[38;5;124m\"\u001B[39m\u001B[38;5;124mbrute\u001B[39m\u001B[38;5;124m\"\u001B[39m\n\u001B[1;32m    758\u001B[0m     \u001B[38;5;129;01mand\u001B[39;00m PairwiseDistancesArgKmin\u001B[38;5;241m.\u001B[39mis_usable_for(\n\u001B[1;32m    759\u001B[0m         X \u001B[38;5;28;01mif\u001B[39;00m X \u001B[38;5;129;01mis\u001B[39;00m \u001B[38;5;129;01mnot\u001B[39;00m \u001B[38;5;28;01mNone\u001B[39;00m \u001B[38;5;28;01melse\u001B[39;00m \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39m_fit_X, \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39m_fit_X, \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39meffective_metric_\n\u001B[1;32m    760\u001B[0m     )\n\u001B[1;32m    761\u001B[0m )\n\u001B[1;32m    762\u001B[0m \u001B[38;5;28;01mif\u001B[39;00m use_pairwise_distances_reductions:\n\u001B[0;32m--> 763\u001B[0m     results \u001B[38;5;241m=\u001B[39m \u001B[43mPairwiseDistancesArgKmin\u001B[49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43mcompute\u001B[49m\u001B[43m(\u001B[49m\n\u001B[1;32m    764\u001B[0m \u001B[43m        \u001B[49m\u001B[43mX\u001B[49m\u001B[38;5;241;43m=\u001B[39;49m\u001B[43mX\u001B[49m\u001B[43m,\u001B[49m\n\u001B[1;32m    765\u001B[0m \u001B[43m        \u001B[49m\u001B[43mY\u001B[49m\u001B[38;5;241;43m=\u001B[39;49m\u001B[38;5;28;43mself\u001B[39;49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43m_fit_X\u001B[49m\u001B[43m,\u001B[49m\n\u001B[1;32m    766\u001B[0m \u001B[43m        \u001B[49m\u001B[43mk\u001B[49m\u001B[38;5;241;43m=\u001B[39;49m\u001B[43mn_neighbors\u001B[49m\u001B[43m,\u001B[49m\n\u001B[1;32m    767\u001B[0m \u001B[43m        \u001B[49m\u001B[43mmetric\u001B[49m\u001B[38;5;241;43m=\u001B[39;49m\u001B[38;5;28;43mself\u001B[39;49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43meffective_metric_\u001B[49m\u001B[43m,\u001B[49m\n\u001B[1;32m    768\u001B[0m \u001B[43m        \u001B[49m\u001B[43mmetric_kwargs\u001B[49m\u001B[38;5;241;43m=\u001B[39;49m\u001B[38;5;28;43mself\u001B[39;49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43meffective_metric_params_\u001B[49m\u001B[43m,\u001B[49m\n\u001B[1;32m    769\u001B[0m \u001B[43m        \u001B[49m\u001B[43mstrategy\u001B[49m\u001B[38;5;241;43m=\u001B[39;49m\u001B[38;5;124;43m\"\u001B[39;49m\u001B[38;5;124;43mauto\u001B[39;49m\u001B[38;5;124;43m\"\u001B[39;49m\u001B[43m,\u001B[49m\n\u001B[1;32m    770\u001B[0m \u001B[43m        \u001B[49m\u001B[43mreturn_distance\u001B[49m\u001B[38;5;241;43m=\u001B[39;49m\u001B[43mreturn_distance\u001B[49m\u001B[43m,\u001B[49m\n\u001B[1;32m    771\u001B[0m \u001B[43m    \u001B[49m\u001B[43m)\u001B[49m\n\u001B[1;32m    773\u001B[0m \u001B[38;5;28;01melif\u001B[39;00m (\n\u001B[1;32m    774\u001B[0m     \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39m_fit_method \u001B[38;5;241m==\u001B[39m \u001B[38;5;124m\"\u001B[39m\u001B[38;5;124mbrute\u001B[39m\u001B[38;5;124m\"\u001B[39m \u001B[38;5;129;01mand\u001B[39;00m \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39mmetric \u001B[38;5;241m==\u001B[39m \u001B[38;5;124m\"\u001B[39m\u001B[38;5;124mprecomputed\u001B[39m\u001B[38;5;124m\"\u001B[39m \u001B[38;5;129;01mand\u001B[39;00m issparse(X)\n\u001B[1;32m    775\u001B[0m ):\n\u001B[1;32m    776\u001B[0m     results \u001B[38;5;241m=\u001B[39m _kneighbors_from_graph(\n\u001B[1;32m    777\u001B[0m         X, n_neighbors\u001B[38;5;241m=\u001B[39mn_neighbors, return_distance\u001B[38;5;241m=\u001B[39mreturn_distance\n\u001B[1;32m    778\u001B[0m     )\n",
      "File \u001B[0;32msklearn/metrics/_pairwise_distances_reduction.pyx:691\u001B[0m, in \u001B[0;36msklearn.metrics._pairwise_distances_reduction.PairwiseDistancesArgKmin.compute\u001B[0;34m()\u001B[0m\n",
      "File \u001B[0;32m~/opt/anaconda3/lib/python3.9/site-packages/sklearn/utils/fixes.py:151\u001B[0m, in \u001B[0;36mthreadpool_limits\u001B[0;34m(limits, user_api)\u001B[0m\n\u001B[1;32m    149\u001B[0m     \u001B[38;5;28;01mreturn\u001B[39;00m controller\u001B[38;5;241m.\u001B[39mlimit(limits\u001B[38;5;241m=\u001B[39mlimits, user_api\u001B[38;5;241m=\u001B[39muser_api)\n\u001B[1;32m    150\u001B[0m \u001B[38;5;28;01melse\u001B[39;00m:\n\u001B[0;32m--> 151\u001B[0m     \u001B[38;5;28;01mreturn\u001B[39;00m \u001B[43mthreadpoolctl\u001B[49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43mthreadpool_limits\u001B[49m\u001B[43m(\u001B[49m\u001B[43mlimits\u001B[49m\u001B[38;5;241;43m=\u001B[39;49m\u001B[43mlimits\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[43muser_api\u001B[49m\u001B[38;5;241;43m=\u001B[39;49m\u001B[43muser_api\u001B[49m\u001B[43m)\u001B[49m\n",
      "File \u001B[0;32m~/opt/anaconda3/lib/python3.9/site-packages/threadpoolctl.py:171\u001B[0m, in \u001B[0;36mthreadpool_limits.__init__\u001B[0;34m(self, limits, user_api)\u001B[0m\n\u001B[1;32m    167\u001B[0m \u001B[38;5;28;01mdef\u001B[39;00m \u001B[38;5;21m__init__\u001B[39m(\u001B[38;5;28mself\u001B[39m, limits\u001B[38;5;241m=\u001B[39m\u001B[38;5;28;01mNone\u001B[39;00m, user_api\u001B[38;5;241m=\u001B[39m\u001B[38;5;28;01mNone\u001B[39;00m):\n\u001B[1;32m    168\u001B[0m     \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39m_limits, \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39m_user_api, \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39m_prefixes \u001B[38;5;241m=\u001B[39m \\\n\u001B[1;32m    169\u001B[0m         \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39m_check_params(limits, user_api)\n\u001B[0;32m--> 171\u001B[0m     \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39m_original_info \u001B[38;5;241m=\u001B[39m \u001B[38;5;28;43mself\u001B[39;49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43m_set_threadpool_limits\u001B[49m\u001B[43m(\u001B[49m\u001B[43m)\u001B[49m\n",
      "File \u001B[0;32m~/opt/anaconda3/lib/python3.9/site-packages/threadpoolctl.py:268\u001B[0m, in \u001B[0;36mthreadpool_limits._set_threadpool_limits\u001B[0;34m(self)\u001B[0m\n\u001B[1;32m    265\u001B[0m \u001B[38;5;28;01mif\u001B[39;00m \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39m_limits \u001B[38;5;129;01mis\u001B[39;00m \u001B[38;5;28;01mNone\u001B[39;00m:\n\u001B[1;32m    266\u001B[0m     \u001B[38;5;28;01mreturn\u001B[39;00m \u001B[38;5;28;01mNone\u001B[39;00m\n\u001B[0;32m--> 268\u001B[0m modules \u001B[38;5;241m=\u001B[39m \u001B[43m_ThreadpoolInfo\u001B[49m\u001B[43m(\u001B[49m\u001B[43mprefixes\u001B[49m\u001B[38;5;241;43m=\u001B[39;49m\u001B[38;5;28;43mself\u001B[39;49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43m_prefixes\u001B[49m\u001B[43m,\u001B[49m\n\u001B[1;32m    269\u001B[0m \u001B[43m                          \u001B[49m\u001B[43muser_api\u001B[49m\u001B[38;5;241;43m=\u001B[39;49m\u001B[38;5;28;43mself\u001B[39;49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43m_user_api\u001B[49m\u001B[43m)\u001B[49m\n\u001B[1;32m    270\u001B[0m \u001B[38;5;28;01mfor\u001B[39;00m module \u001B[38;5;129;01min\u001B[39;00m modules:\n\u001B[1;32m    271\u001B[0m     \u001B[38;5;66;03m# self._limits is a dict {key: num_threads} where key is either\u001B[39;00m\n\u001B[1;32m    272\u001B[0m     \u001B[38;5;66;03m# a prefix or a user_api. If a module matches both, the limit\u001B[39;00m\n\u001B[1;32m    273\u001B[0m     \u001B[38;5;66;03m# corresponding to the prefix is chosed.\u001B[39;00m\n\u001B[1;32m    274\u001B[0m     \u001B[38;5;28;01mif\u001B[39;00m module\u001B[38;5;241m.\u001B[39mprefix \u001B[38;5;129;01min\u001B[39;00m \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39m_limits:\n",
      "File \u001B[0;32m~/opt/anaconda3/lib/python3.9/site-packages/threadpoolctl.py:340\u001B[0m, in \u001B[0;36m_ThreadpoolInfo.__init__\u001B[0;34m(self, user_api, prefixes, modules)\u001B[0m\n\u001B[1;32m    337\u001B[0m     \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39muser_api \u001B[38;5;241m=\u001B[39m [] \u001B[38;5;28;01mif\u001B[39;00m user_api \u001B[38;5;129;01mis\u001B[39;00m \u001B[38;5;28;01mNone\u001B[39;00m \u001B[38;5;28;01melse\u001B[39;00m user_api\n\u001B[1;32m    339\u001B[0m     \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39mmodules \u001B[38;5;241m=\u001B[39m []\n\u001B[0;32m--> 340\u001B[0m     \u001B[38;5;28;43mself\u001B[39;49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43m_load_modules\u001B[49m\u001B[43m(\u001B[49m\u001B[43m)\u001B[49m\n\u001B[1;32m    341\u001B[0m     \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39m_warn_if_incompatible_openmp()\n\u001B[1;32m    342\u001B[0m \u001B[38;5;28;01melse\u001B[39;00m:\n",
      "File \u001B[0;32m~/opt/anaconda3/lib/python3.9/site-packages/threadpoolctl.py:371\u001B[0m, in \u001B[0;36m_ThreadpoolInfo._load_modules\u001B[0;34m(self)\u001B[0m\n\u001B[1;32m    369\u001B[0m \u001B[38;5;124;03m\"\"\"Loop through loaded libraries and store supported ones\"\"\"\u001B[39;00m\n\u001B[1;32m    370\u001B[0m \u001B[38;5;28;01mif\u001B[39;00m sys\u001B[38;5;241m.\u001B[39mplatform \u001B[38;5;241m==\u001B[39m \u001B[38;5;124m\"\u001B[39m\u001B[38;5;124mdarwin\u001B[39m\u001B[38;5;124m\"\u001B[39m:\n\u001B[0;32m--> 371\u001B[0m     \u001B[38;5;28;43mself\u001B[39;49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43m_find_modules_with_dyld\u001B[49m\u001B[43m(\u001B[49m\u001B[43m)\u001B[49m\n\u001B[1;32m    372\u001B[0m \u001B[38;5;28;01melif\u001B[39;00m sys\u001B[38;5;241m.\u001B[39mplatform \u001B[38;5;241m==\u001B[39m \u001B[38;5;124m\"\u001B[39m\u001B[38;5;124mwin32\u001B[39m\u001B[38;5;124m\"\u001B[39m:\n\u001B[1;32m    373\u001B[0m     \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39m_find_modules_with_enum_process_module_ex()\n",
      "File \u001B[0;32m~/opt/anaconda3/lib/python3.9/site-packages/threadpoolctl.py:428\u001B[0m, in \u001B[0;36m_ThreadpoolInfo._find_modules_with_dyld\u001B[0;34m(self)\u001B[0m\n\u001B[1;32m    425\u001B[0m filepath \u001B[38;5;241m=\u001B[39m filepath\u001B[38;5;241m.\u001B[39mdecode(\u001B[38;5;124m\"\u001B[39m\u001B[38;5;124mutf-8\u001B[39m\u001B[38;5;124m\"\u001B[39m)\n\u001B[1;32m    427\u001B[0m \u001B[38;5;66;03m# Store the module if it is supported and selected\u001B[39;00m\n\u001B[0;32m--> 428\u001B[0m \u001B[38;5;28;43mself\u001B[39;49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43m_make_module_from_path\u001B[49m\u001B[43m(\u001B[49m\u001B[43mfilepath\u001B[49m\u001B[43m)\u001B[49m\n",
      "File \u001B[0;32m~/opt/anaconda3/lib/python3.9/site-packages/threadpoolctl.py:515\u001B[0m, in \u001B[0;36m_ThreadpoolInfo._make_module_from_path\u001B[0;34m(self, filepath)\u001B[0m\n\u001B[1;32m    513\u001B[0m \u001B[38;5;28;01mif\u001B[39;00m prefix \u001B[38;5;129;01min\u001B[39;00m \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39mprefixes \u001B[38;5;129;01mor\u001B[39;00m user_api \u001B[38;5;129;01min\u001B[39;00m \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39muser_api:\n\u001B[1;32m    514\u001B[0m     module_class \u001B[38;5;241m=\u001B[39m \u001B[38;5;28mglobals\u001B[39m()[module_class]\n\u001B[0;32m--> 515\u001B[0m     module \u001B[38;5;241m=\u001B[39m \u001B[43mmodule_class\u001B[49m\u001B[43m(\u001B[49m\u001B[43mfilepath\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[43mprefix\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[43muser_api\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[43minternal_api\u001B[49m\u001B[43m)\u001B[49m\n\u001B[1;32m    516\u001B[0m     \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39mmodules\u001B[38;5;241m.\u001B[39mappend(module)\n",
      "File \u001B[0;32m~/opt/anaconda3/lib/python3.9/site-packages/threadpoolctl.py:608\u001B[0m, in \u001B[0;36m_Module.__init__\u001B[0;34m(self, filepath, prefix, user_api, internal_api)\u001B[0m\n\u001B[1;32m    606\u001B[0m \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39mversion \u001B[38;5;241m=\u001B[39m \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39mget_version()\n\u001B[1;32m    607\u001B[0m \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39mnum_threads \u001B[38;5;241m=\u001B[39m \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39mget_num_threads()\n\u001B[0;32m--> 608\u001B[0m \u001B[38;5;28;43mself\u001B[39;49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43m_get_extra_info\u001B[49m\u001B[43m(\u001B[49m\u001B[43m)\u001B[49m\n",
      "File \u001B[0;32m~/opt/anaconda3/lib/python3.9/site-packages/threadpoolctl.py:662\u001B[0m, in \u001B[0;36m_OpenBLASModule._get_extra_info\u001B[0;34m(self)\u001B[0m\n\u001B[1;32m    661\u001B[0m \u001B[38;5;28;01mdef\u001B[39;00m \u001B[38;5;21m_get_extra_info\u001B[39m(\u001B[38;5;28mself\u001B[39m):\n\u001B[0;32m--> 662\u001B[0m     \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39mthreading_layer \u001B[38;5;241m=\u001B[39m \u001B[38;5;28;43mself\u001B[39;49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43mget_threading_layer\u001B[49m\u001B[43m(\u001B[49m\u001B[43m)\u001B[49m\n\u001B[1;32m    663\u001B[0m     \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39marchitecture \u001B[38;5;241m=\u001B[39m \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39mget_architecture()\n",
      "File \u001B[0;32m~/opt/anaconda3/lib/python3.9/site-packages/threadpoolctl.py:667\u001B[0m, in \u001B[0;36m_OpenBLASModule.get_threading_layer\u001B[0;34m(self)\u001B[0m\n\u001B[1;32m    665\u001B[0m \u001B[38;5;28;01mdef\u001B[39;00m \u001B[38;5;21mget_threading_layer\u001B[39m(\u001B[38;5;28mself\u001B[39m):\n\u001B[1;32m    666\u001B[0m     \u001B[38;5;124;03m\"\"\"Return the threading layer of OpenBLAS\"\"\"\u001B[39;00m\n\u001B[0;32m--> 667\u001B[0m     threading_layer \u001B[38;5;241m=\u001B[39m \u001B[38;5;28;43mself\u001B[39;49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43m_dynlib\u001B[49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43mopenblas_get_parallel\u001B[49m()\n\u001B[1;32m    668\u001B[0m     \u001B[38;5;28;01mif\u001B[39;00m threading_layer \u001B[38;5;241m==\u001B[39m \u001B[38;5;241m2\u001B[39m:\n\u001B[1;32m    669\u001B[0m         \u001B[38;5;28;01mreturn\u001B[39;00m \u001B[38;5;124m\"\u001B[39m\u001B[38;5;124mopenmp\u001B[39m\u001B[38;5;124m\"\u001B[39m\n",
      "File \u001B[0;32m~/opt/anaconda3/lib/python3.9/ctypes/__init__.py:395\u001B[0m, in \u001B[0;36mCDLL.__getattr__\u001B[0;34m(self, name)\u001B[0m\n\u001B[1;32m    393\u001B[0m \u001B[38;5;28;01mif\u001B[39;00m name\u001B[38;5;241m.\u001B[39mstartswith(\u001B[38;5;124m'\u001B[39m\u001B[38;5;124m__\u001B[39m\u001B[38;5;124m'\u001B[39m) \u001B[38;5;129;01mand\u001B[39;00m name\u001B[38;5;241m.\u001B[39mendswith(\u001B[38;5;124m'\u001B[39m\u001B[38;5;124m__\u001B[39m\u001B[38;5;124m'\u001B[39m):\n\u001B[1;32m    394\u001B[0m     \u001B[38;5;28;01mraise\u001B[39;00m \u001B[38;5;167;01mAttributeError\u001B[39;00m(name)\n\u001B[0;32m--> 395\u001B[0m func \u001B[38;5;241m=\u001B[39m \u001B[38;5;28;43mself\u001B[39;49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[38;5;21;43m__getitem__\u001B[39;49m\u001B[43m(\u001B[49m\u001B[43mname\u001B[49m\u001B[43m)\u001B[49m\n\u001B[1;32m    396\u001B[0m \u001B[38;5;28msetattr\u001B[39m(\u001B[38;5;28mself\u001B[39m, name, func)\n\u001B[1;32m    397\u001B[0m \u001B[38;5;28;01mreturn\u001B[39;00m func\n",
      "File \u001B[0;32m~/opt/anaconda3/lib/python3.9/ctypes/__init__.py:400\u001B[0m, in \u001B[0;36mCDLL.__getitem__\u001B[0;34m(self, name_or_ordinal)\u001B[0m\n\u001B[1;32m    399\u001B[0m \u001B[38;5;28;01mdef\u001B[39;00m \u001B[38;5;21m__getitem__\u001B[39m(\u001B[38;5;28mself\u001B[39m, name_or_ordinal):\n\u001B[0;32m--> 400\u001B[0m     func \u001B[38;5;241m=\u001B[39m \u001B[38;5;28;43mself\u001B[39;49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43m_FuncPtr\u001B[49m\u001B[43m(\u001B[49m\u001B[43m(\u001B[49m\u001B[43mname_or_ordinal\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[38;5;28;43mself\u001B[39;49m\u001B[43m)\u001B[49m\u001B[43m)\u001B[49m\n\u001B[1;32m    401\u001B[0m     \u001B[38;5;28;01mif\u001B[39;00m \u001B[38;5;129;01mnot\u001B[39;00m \u001B[38;5;28misinstance\u001B[39m(name_or_ordinal, \u001B[38;5;28mint\u001B[39m):\n\u001B[1;32m    402\u001B[0m         func\u001B[38;5;241m.\u001B[39m\u001B[38;5;18m__name__\u001B[39m \u001B[38;5;241m=\u001B[39m name_or_ordinal\n",
      "\u001B[0;31mAttributeError\u001B[0m: dlsym(0x9ade32b0, openblas_get_parallel): symbol not found"
     ]
    }
   ],
   "source": [
    "# KNN模型 KNeighborsClassifier\n",
    "# 数据2：分类\n",
    "\n",
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "from sklearn.metrics import mean_squared_error\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "\n",
    "stdScaler = StandardScaler()\n",
    "X = stdScaler.fit_transform(train)\n",
    "# Split the data into a training set and a test set\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, target, random_state=0)\n",
    "clf = KNeighborsClassifier(n_neighbors=3)\n",
    "%time clf.fit(X_train, y_train)\n",
    "clf.score(X_test, y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "27a34ad0",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 3.79 ms, sys: 1.55 ms, total: 5.33 ms\n",
      "Wall time: 5.37 ms\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.424"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# NB模型 GaussianNB\n",
    "# 数据2：分类\n",
    "\n",
    "from sklearn.naive_bayes import GaussianNB\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "stdScaler = StandardScaler()\n",
    "X = stdScaler.fit_transform(train)\n",
    "# Split the data into a training set and a test set\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, target, random_state=0)\n",
    "%time clf = GaussianNB().fit(X_train, y_train)\n",
    "clf.score(X_test, y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "ebeb61f4",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 77.9 ms, sys: 400 µs, total: 78.3 ms\n",
      "Wall time: 78.6 ms\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.864"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 树模型 tree\n",
    "# 数据2：分类\n",
    "\n",
    "from sklearn import tree\n",
    "# Split the data into a training set and a test set\n",
    "X_train, X_test, y_train, y_test = train_test_split(train, target, random_state=0)\n",
    "clf = tree.DecisionTreeClassifier()\n",
    "%time clf = clf.fit(X_train, y_train)\n",
    "clf.score(X_test, y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "10dc8f2f",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 14.7 ms, sys: 1.18 ms, total: 15.9 ms\n",
      "Wall time: 15.9 ms\n"
     ]
    },
    {
     "ename": "AttributeError",
     "evalue": "dlsym(0x9ade32b0, openblas_get_parallel): symbol not found",
     "output_type": "error",
     "traceback": [
      "\u001B[0;31m---------------------------------------------------------------------------\u001B[0m",
      "\u001B[0;31mAttributeError\u001B[0m                            Traceback (most recent call last)",
      "Input \u001B[0;32mIn [9]\u001B[0m, in \u001B[0;36m<cell line: 10>\u001B[0;34m()\u001B[0m\n\u001B[1;32m      8\u001B[0m clf \u001B[38;5;241m=\u001B[39m BaggingClassifier(KNeighborsClassifier(), max_samples\u001B[38;5;241m=\u001B[39m\u001B[38;5;241m0.5\u001B[39m, max_features\u001B[38;5;241m=\u001B[39m\u001B[38;5;241m0.5\u001B[39m)\n\u001B[1;32m      9\u001B[0m get_ipython()\u001B[38;5;241m.\u001B[39mrun_line_magic(\u001B[38;5;124m'\u001B[39m\u001B[38;5;124mtime\u001B[39m\u001B[38;5;124m'\u001B[39m, \u001B[38;5;124m'\u001B[39m\u001B[38;5;124mclf = clf.fit(X_train, y_train)\u001B[39m\u001B[38;5;124m'\u001B[39m)\n\u001B[0;32m---> 10\u001B[0m \u001B[43mclf\u001B[49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43mscore\u001B[49m\u001B[43m(\u001B[49m\u001B[43mX_test\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[43my_test\u001B[49m\u001B[43m)\u001B[49m\n",
      "File \u001B[0;32m~/opt/anaconda3/lib/python3.9/site-packages/sklearn/base.py:666\u001B[0m, in \u001B[0;36mClassifierMixin.score\u001B[0;34m(self, X, y, sample_weight)\u001B[0m\n\u001B[1;32m    641\u001B[0m \u001B[38;5;124;03m\"\"\"\u001B[39;00m\n\u001B[1;32m    642\u001B[0m \u001B[38;5;124;03mReturn the mean accuracy on the given test data and labels.\u001B[39;00m\n\u001B[1;32m    643\u001B[0m \n\u001B[0;32m   (...)\u001B[0m\n\u001B[1;32m    662\u001B[0m \u001B[38;5;124;03m    Mean accuracy of ``self.predict(X)`` wrt. `y`.\u001B[39;00m\n\u001B[1;32m    663\u001B[0m \u001B[38;5;124;03m\"\"\"\u001B[39;00m\n\u001B[1;32m    664\u001B[0m \u001B[38;5;28;01mfrom\u001B[39;00m \u001B[38;5;21;01m.\u001B[39;00m\u001B[38;5;21;01mmetrics\u001B[39;00m \u001B[38;5;28;01mimport\u001B[39;00m accuracy_score\n\u001B[0;32m--> 666\u001B[0m \u001B[38;5;28;01mreturn\u001B[39;00m accuracy_score(y, \u001B[38;5;28;43mself\u001B[39;49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43mpredict\u001B[49m\u001B[43m(\u001B[49m\u001B[43mX\u001B[49m\u001B[43m)\u001B[49m, sample_weight\u001B[38;5;241m=\u001B[39msample_weight)\n",
      "File \u001B[0;32m~/opt/anaconda3/lib/python3.9/site-packages/sklearn/ensemble/_bagging.py:785\u001B[0m, in \u001B[0;36mBaggingClassifier.predict\u001B[0;34m(self, X)\u001B[0m\n\u001B[1;32m    767\u001B[0m \u001B[38;5;28;01mdef\u001B[39;00m \u001B[38;5;21mpredict\u001B[39m(\u001B[38;5;28mself\u001B[39m, X):\n\u001B[1;32m    768\u001B[0m     \u001B[38;5;124;03m\"\"\"Predict class for X.\u001B[39;00m\n\u001B[1;32m    769\u001B[0m \n\u001B[1;32m    770\u001B[0m \u001B[38;5;124;03m    The predicted class of an input sample is computed as the class with\u001B[39;00m\n\u001B[0;32m   (...)\u001B[0m\n\u001B[1;32m    783\u001B[0m \u001B[38;5;124;03m        The predicted classes.\u001B[39;00m\n\u001B[1;32m    784\u001B[0m \u001B[38;5;124;03m    \"\"\"\u001B[39;00m\n\u001B[0;32m--> 785\u001B[0m     predicted_probabilitiy \u001B[38;5;241m=\u001B[39m \u001B[38;5;28;43mself\u001B[39;49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43mpredict_proba\u001B[49m\u001B[43m(\u001B[49m\u001B[43mX\u001B[49m\u001B[43m)\u001B[49m\n\u001B[1;32m    786\u001B[0m     \u001B[38;5;28;01mreturn\u001B[39;00m \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39mclasses_\u001B[38;5;241m.\u001B[39mtake((np\u001B[38;5;241m.\u001B[39margmax(predicted_probabilitiy, axis\u001B[38;5;241m=\u001B[39m\u001B[38;5;241m1\u001B[39m)), axis\u001B[38;5;241m=\u001B[39m\u001B[38;5;241m0\u001B[39m)\n",
      "File \u001B[0;32m~/opt/anaconda3/lib/python3.9/site-packages/sklearn/ensemble/_bagging.py:825\u001B[0m, in \u001B[0;36mBaggingClassifier.predict_proba\u001B[0;34m(self, X)\u001B[0m\n\u001B[1;32m    820\u001B[0m \u001B[38;5;66;03m# Parallel loop\u001B[39;00m\n\u001B[1;32m    821\u001B[0m n_jobs, n_estimators, starts \u001B[38;5;241m=\u001B[39m _partition_estimators(\n\u001B[1;32m    822\u001B[0m     \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39mn_estimators, \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39mn_jobs\n\u001B[1;32m    823\u001B[0m )\n\u001B[0;32m--> 825\u001B[0m all_proba \u001B[38;5;241m=\u001B[39m \u001B[43mParallel\u001B[49m\u001B[43m(\u001B[49m\n\u001B[1;32m    826\u001B[0m \u001B[43m    \u001B[49m\u001B[43mn_jobs\u001B[49m\u001B[38;5;241;43m=\u001B[39;49m\u001B[43mn_jobs\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[43mverbose\u001B[49m\u001B[38;5;241;43m=\u001B[39;49m\u001B[38;5;28;43mself\u001B[39;49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43mverbose\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[38;5;241;43m*\u001B[39;49m\u001B[38;5;241;43m*\u001B[39;49m\u001B[38;5;28;43mself\u001B[39;49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43m_parallel_args\u001B[49m\u001B[43m(\u001B[49m\u001B[43m)\u001B[49m\n\u001B[1;32m    827\u001B[0m \u001B[43m\u001B[49m\u001B[43m)\u001B[49m\u001B[43m(\u001B[49m\n\u001B[1;32m    828\u001B[0m \u001B[43m    \u001B[49m\u001B[43mdelayed\u001B[49m\u001B[43m(\u001B[49m\u001B[43m_parallel_predict_proba\u001B[49m\u001B[43m)\u001B[49m\u001B[43m(\u001B[49m\n\u001B[1;32m    829\u001B[0m \u001B[43m        \u001B[49m\u001B[38;5;28;43mself\u001B[39;49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43mestimators_\u001B[49m\u001B[43m[\u001B[49m\u001B[43mstarts\u001B[49m\u001B[43m[\u001B[49m\u001B[43mi\u001B[49m\u001B[43m]\u001B[49m\u001B[43m \u001B[49m\u001B[43m:\u001B[49m\u001B[43m \u001B[49m\u001B[43mstarts\u001B[49m\u001B[43m[\u001B[49m\u001B[43mi\u001B[49m\u001B[43m \u001B[49m\u001B[38;5;241;43m+\u001B[39;49m\u001B[43m \u001B[49m\u001B[38;5;241;43m1\u001B[39;49m\u001B[43m]\u001B[49m\u001B[43m]\u001B[49m\u001B[43m,\u001B[49m\n\u001B[1;32m    830\u001B[0m \u001B[43m        \u001B[49m\u001B[38;5;28;43mself\u001B[39;49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43mestimators_features_\u001B[49m\u001B[43m[\u001B[49m\u001B[43mstarts\u001B[49m\u001B[43m[\u001B[49m\u001B[43mi\u001B[49m\u001B[43m]\u001B[49m\u001B[43m \u001B[49m\u001B[43m:\u001B[49m\u001B[43m \u001B[49m\u001B[43mstarts\u001B[49m\u001B[43m[\u001B[49m\u001B[43mi\u001B[49m\u001B[43m \u001B[49m\u001B[38;5;241;43m+\u001B[39;49m\u001B[43m \u001B[49m\u001B[38;5;241;43m1\u001B[39;49m\u001B[43m]\u001B[49m\u001B[43m]\u001B[49m\u001B[43m,\u001B[49m\n\u001B[1;32m    831\u001B[0m \u001B[43m        \u001B[49m\u001B[43mX\u001B[49m\u001B[43m,\u001B[49m\n\u001B[1;32m    832\u001B[0m \u001B[43m        \u001B[49m\u001B[38;5;28;43mself\u001B[39;49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43mn_classes_\u001B[49m\u001B[43m,\u001B[49m\n\u001B[1;32m    833\u001B[0m \u001B[43m    \u001B[49m\u001B[43m)\u001B[49m\n\u001B[1;32m    834\u001B[0m \u001B[43m    \u001B[49m\u001B[38;5;28;43;01mfor\u001B[39;49;00m\u001B[43m \u001B[49m\u001B[43mi\u001B[49m\u001B[43m \u001B[49m\u001B[38;5;129;43;01min\u001B[39;49;00m\u001B[43m \u001B[49m\u001B[38;5;28;43mrange\u001B[39;49m\u001B[43m(\u001B[49m\u001B[43mn_jobs\u001B[49m\u001B[43m)\u001B[49m\n\u001B[1;32m    835\u001B[0m \u001B[43m\u001B[49m\u001B[43m)\u001B[49m\n\u001B[1;32m    837\u001B[0m \u001B[38;5;66;03m# Reduce\u001B[39;00m\n\u001B[1;32m    838\u001B[0m proba \u001B[38;5;241m=\u001B[39m \u001B[38;5;28msum\u001B[39m(all_proba) \u001B[38;5;241m/\u001B[39m \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39mn_estimators\n",
      "File \u001B[0;32m~/opt/anaconda3/lib/python3.9/site-packages/joblib/parallel.py:1043\u001B[0m, in \u001B[0;36mParallel.__call__\u001B[0;34m(self, iterable)\u001B[0m\n\u001B[1;32m   1034\u001B[0m \u001B[38;5;28;01mtry\u001B[39;00m:\n\u001B[1;32m   1035\u001B[0m     \u001B[38;5;66;03m# Only set self._iterating to True if at least a batch\u001B[39;00m\n\u001B[1;32m   1036\u001B[0m     \u001B[38;5;66;03m# was dispatched. In particular this covers the edge\u001B[39;00m\n\u001B[0;32m   (...)\u001B[0m\n\u001B[1;32m   1040\u001B[0m     \u001B[38;5;66;03m# was very quick and its callback already dispatched all the\u001B[39;00m\n\u001B[1;32m   1041\u001B[0m     \u001B[38;5;66;03m# remaining jobs.\u001B[39;00m\n\u001B[1;32m   1042\u001B[0m     \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39m_iterating \u001B[38;5;241m=\u001B[39m \u001B[38;5;28;01mFalse\u001B[39;00m\n\u001B[0;32m-> 1043\u001B[0m     \u001B[38;5;28;01mif\u001B[39;00m \u001B[38;5;28;43mself\u001B[39;49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43mdispatch_one_batch\u001B[49m\u001B[43m(\u001B[49m\u001B[43miterator\u001B[49m\u001B[43m)\u001B[49m:\n\u001B[1;32m   1044\u001B[0m         \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39m_iterating \u001B[38;5;241m=\u001B[39m \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39m_original_iterator \u001B[38;5;129;01mis\u001B[39;00m \u001B[38;5;129;01mnot\u001B[39;00m \u001B[38;5;28;01mNone\u001B[39;00m\n\u001B[1;32m   1046\u001B[0m     \u001B[38;5;28;01mwhile\u001B[39;00m \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39mdispatch_one_batch(iterator):\n",
      "File \u001B[0;32m~/opt/anaconda3/lib/python3.9/site-packages/joblib/parallel.py:861\u001B[0m, in \u001B[0;36mParallel.dispatch_one_batch\u001B[0;34m(self, iterator)\u001B[0m\n\u001B[1;32m    859\u001B[0m     \u001B[38;5;28;01mreturn\u001B[39;00m \u001B[38;5;28;01mFalse\u001B[39;00m\n\u001B[1;32m    860\u001B[0m \u001B[38;5;28;01melse\u001B[39;00m:\n\u001B[0;32m--> 861\u001B[0m     \u001B[38;5;28;43mself\u001B[39;49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43m_dispatch\u001B[49m\u001B[43m(\u001B[49m\u001B[43mtasks\u001B[49m\u001B[43m)\u001B[49m\n\u001B[1;32m    862\u001B[0m     \u001B[38;5;28;01mreturn\u001B[39;00m \u001B[38;5;28;01mTrue\u001B[39;00m\n",
      "File \u001B[0;32m~/opt/anaconda3/lib/python3.9/site-packages/joblib/parallel.py:779\u001B[0m, in \u001B[0;36mParallel._dispatch\u001B[0;34m(self, batch)\u001B[0m\n\u001B[1;32m    777\u001B[0m \u001B[38;5;28;01mwith\u001B[39;00m \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39m_lock:\n\u001B[1;32m    778\u001B[0m     job_idx \u001B[38;5;241m=\u001B[39m \u001B[38;5;28mlen\u001B[39m(\u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39m_jobs)\n\u001B[0;32m--> 779\u001B[0m     job \u001B[38;5;241m=\u001B[39m \u001B[38;5;28;43mself\u001B[39;49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43m_backend\u001B[49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43mapply_async\u001B[49m\u001B[43m(\u001B[49m\u001B[43mbatch\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[43mcallback\u001B[49m\u001B[38;5;241;43m=\u001B[39;49m\u001B[43mcb\u001B[49m\u001B[43m)\u001B[49m\n\u001B[1;32m    780\u001B[0m     \u001B[38;5;66;03m# A job can complete so quickly than its callback is\u001B[39;00m\n\u001B[1;32m    781\u001B[0m     \u001B[38;5;66;03m# called before we get here, causing self._jobs to\u001B[39;00m\n\u001B[1;32m    782\u001B[0m     \u001B[38;5;66;03m# grow. To ensure correct results ordering, .insert is\u001B[39;00m\n\u001B[1;32m    783\u001B[0m     \u001B[38;5;66;03m# used (rather than .append) in the following line\u001B[39;00m\n\u001B[1;32m    784\u001B[0m     \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39m_jobs\u001B[38;5;241m.\u001B[39minsert(job_idx, job)\n",
      "File \u001B[0;32m~/opt/anaconda3/lib/python3.9/site-packages/joblib/_parallel_backends.py:208\u001B[0m, in \u001B[0;36mSequentialBackend.apply_async\u001B[0;34m(self, func, callback)\u001B[0m\n\u001B[1;32m    206\u001B[0m \u001B[38;5;28;01mdef\u001B[39;00m \u001B[38;5;21mapply_async\u001B[39m(\u001B[38;5;28mself\u001B[39m, func, callback\u001B[38;5;241m=\u001B[39m\u001B[38;5;28;01mNone\u001B[39;00m):\n\u001B[1;32m    207\u001B[0m     \u001B[38;5;124;03m\"\"\"Schedule a func to be run\"\"\"\u001B[39;00m\n\u001B[0;32m--> 208\u001B[0m     result \u001B[38;5;241m=\u001B[39m \u001B[43mImmediateResult\u001B[49m\u001B[43m(\u001B[49m\u001B[43mfunc\u001B[49m\u001B[43m)\u001B[49m\n\u001B[1;32m    209\u001B[0m     \u001B[38;5;28;01mif\u001B[39;00m callback:\n\u001B[1;32m    210\u001B[0m         callback(result)\n",
      "File \u001B[0;32m~/opt/anaconda3/lib/python3.9/site-packages/joblib/_parallel_backends.py:572\u001B[0m, in \u001B[0;36mImmediateResult.__init__\u001B[0;34m(self, batch)\u001B[0m\n\u001B[1;32m    569\u001B[0m \u001B[38;5;28;01mdef\u001B[39;00m \u001B[38;5;21m__init__\u001B[39m(\u001B[38;5;28mself\u001B[39m, batch):\n\u001B[1;32m    570\u001B[0m     \u001B[38;5;66;03m# Don't delay the application, to avoid keeping the input\u001B[39;00m\n\u001B[1;32m    571\u001B[0m     \u001B[38;5;66;03m# arguments in memory\u001B[39;00m\n\u001B[0;32m--> 572\u001B[0m     \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39mresults \u001B[38;5;241m=\u001B[39m \u001B[43mbatch\u001B[49m\u001B[43m(\u001B[49m\u001B[43m)\u001B[49m\n",
      "File \u001B[0;32m~/opt/anaconda3/lib/python3.9/site-packages/joblib/parallel.py:262\u001B[0m, in \u001B[0;36mBatchedCalls.__call__\u001B[0;34m(self)\u001B[0m\n\u001B[1;32m    258\u001B[0m \u001B[38;5;28;01mdef\u001B[39;00m \u001B[38;5;21m__call__\u001B[39m(\u001B[38;5;28mself\u001B[39m):\n\u001B[1;32m    259\u001B[0m     \u001B[38;5;66;03m# Set the default nested backend to self._backend but do not set the\u001B[39;00m\n\u001B[1;32m    260\u001B[0m     \u001B[38;5;66;03m# change the default number of processes to -1\u001B[39;00m\n\u001B[1;32m    261\u001B[0m     \u001B[38;5;28;01mwith\u001B[39;00m parallel_backend(\u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39m_backend, n_jobs\u001B[38;5;241m=\u001B[39m\u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39m_n_jobs):\n\u001B[0;32m--> 262\u001B[0m         \u001B[38;5;28;01mreturn\u001B[39;00m [func(\u001B[38;5;241m*\u001B[39margs, \u001B[38;5;241m*\u001B[39m\u001B[38;5;241m*\u001B[39mkwargs)\n\u001B[1;32m    263\u001B[0m                 \u001B[38;5;28;01mfor\u001B[39;00m func, args, kwargs \u001B[38;5;129;01min\u001B[39;00m \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39mitems]\n",
      "File \u001B[0;32m~/opt/anaconda3/lib/python3.9/site-packages/joblib/parallel.py:262\u001B[0m, in \u001B[0;36m<listcomp>\u001B[0;34m(.0)\u001B[0m\n\u001B[1;32m    258\u001B[0m \u001B[38;5;28;01mdef\u001B[39;00m \u001B[38;5;21m__call__\u001B[39m(\u001B[38;5;28mself\u001B[39m):\n\u001B[1;32m    259\u001B[0m     \u001B[38;5;66;03m# Set the default nested backend to self._backend but do not set the\u001B[39;00m\n\u001B[1;32m    260\u001B[0m     \u001B[38;5;66;03m# change the default number of processes to -1\u001B[39;00m\n\u001B[1;32m    261\u001B[0m     \u001B[38;5;28;01mwith\u001B[39;00m parallel_backend(\u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39m_backend, n_jobs\u001B[38;5;241m=\u001B[39m\u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39m_n_jobs):\n\u001B[0;32m--> 262\u001B[0m         \u001B[38;5;28;01mreturn\u001B[39;00m [\u001B[43mfunc\u001B[49m\u001B[43m(\u001B[49m\u001B[38;5;241;43m*\u001B[39;49m\u001B[43margs\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[38;5;241;43m*\u001B[39;49m\u001B[38;5;241;43m*\u001B[39;49m\u001B[43mkwargs\u001B[49m\u001B[43m)\u001B[49m\n\u001B[1;32m    263\u001B[0m                 \u001B[38;5;28;01mfor\u001B[39;00m func, args, kwargs \u001B[38;5;129;01min\u001B[39;00m \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39mitems]\n",
      "File \u001B[0;32m~/opt/anaconda3/lib/python3.9/site-packages/sklearn/utils/fixes.py:117\u001B[0m, in \u001B[0;36m_FuncWrapper.__call__\u001B[0;34m(self, *args, **kwargs)\u001B[0m\n\u001B[1;32m    115\u001B[0m \u001B[38;5;28;01mdef\u001B[39;00m \u001B[38;5;21m__call__\u001B[39m(\u001B[38;5;28mself\u001B[39m, \u001B[38;5;241m*\u001B[39margs, \u001B[38;5;241m*\u001B[39m\u001B[38;5;241m*\u001B[39mkwargs):\n\u001B[1;32m    116\u001B[0m     \u001B[38;5;28;01mwith\u001B[39;00m config_context(\u001B[38;5;241m*\u001B[39m\u001B[38;5;241m*\u001B[39m\u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39mconfig):\n\u001B[0;32m--> 117\u001B[0m         \u001B[38;5;28;01mreturn\u001B[39;00m \u001B[38;5;28;43mself\u001B[39;49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43mfunction\u001B[49m\u001B[43m(\u001B[49m\u001B[38;5;241;43m*\u001B[39;49m\u001B[43margs\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[38;5;241;43m*\u001B[39;49m\u001B[38;5;241;43m*\u001B[39;49m\u001B[43mkwargs\u001B[49m\u001B[43m)\u001B[49m\n",
      "File \u001B[0;32m~/opt/anaconda3/lib/python3.9/site-packages/sklearn/ensemble/_bagging.py:156\u001B[0m, in \u001B[0;36m_parallel_predict_proba\u001B[0;34m(estimators, estimators_features, X, n_classes)\u001B[0m\n\u001B[1;32m    154\u001B[0m \u001B[38;5;28;01mfor\u001B[39;00m estimator, features \u001B[38;5;129;01min\u001B[39;00m \u001B[38;5;28mzip\u001B[39m(estimators, estimators_features):\n\u001B[1;32m    155\u001B[0m     \u001B[38;5;28;01mif\u001B[39;00m \u001B[38;5;28mhasattr\u001B[39m(estimator, \u001B[38;5;124m\"\u001B[39m\u001B[38;5;124mpredict_proba\u001B[39m\u001B[38;5;124m\"\u001B[39m):\n\u001B[0;32m--> 156\u001B[0m         proba_estimator \u001B[38;5;241m=\u001B[39m \u001B[43mestimator\u001B[49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43mpredict_proba\u001B[49m\u001B[43m(\u001B[49m\u001B[43mX\u001B[49m\u001B[43m[\u001B[49m\u001B[43m:\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[43mfeatures\u001B[49m\u001B[43m]\u001B[49m\u001B[43m)\u001B[49m\n\u001B[1;32m    158\u001B[0m         \u001B[38;5;28;01mif\u001B[39;00m n_classes \u001B[38;5;241m==\u001B[39m \u001B[38;5;28mlen\u001B[39m(estimator\u001B[38;5;241m.\u001B[39mclasses_):\n\u001B[1;32m    159\u001B[0m             proba \u001B[38;5;241m+\u001B[39m\u001B[38;5;241m=\u001B[39m proba_estimator\n",
      "File \u001B[0;32m~/opt/anaconda3/lib/python3.9/site-packages/sklearn/neighbors/_classification.py:268\u001B[0m, in \u001B[0;36mKNeighborsClassifier.predict_proba\u001B[0;34m(self, X)\u001B[0m\n\u001B[1;32m    250\u001B[0m \u001B[38;5;124;03m\"\"\"Return probability estimates for the test data X.\u001B[39;00m\n\u001B[1;32m    251\u001B[0m \n\u001B[1;32m    252\u001B[0m \u001B[38;5;124;03mParameters\u001B[39;00m\n\u001B[0;32m   (...)\u001B[0m\n\u001B[1;32m    263\u001B[0m \u001B[38;5;124;03m    by lexicographic order.\u001B[39;00m\n\u001B[1;32m    264\u001B[0m \u001B[38;5;124;03m\"\"\"\u001B[39;00m\n\u001B[1;32m    265\u001B[0m \u001B[38;5;28;01mif\u001B[39;00m \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39mweights \u001B[38;5;241m==\u001B[39m \u001B[38;5;124m\"\u001B[39m\u001B[38;5;124muniform\u001B[39m\u001B[38;5;124m\"\u001B[39m:\n\u001B[1;32m    266\u001B[0m     \u001B[38;5;66;03m# In that case, we do not need the distances to perform\u001B[39;00m\n\u001B[1;32m    267\u001B[0m     \u001B[38;5;66;03m# the weighting so we do not compute them.\u001B[39;00m\n\u001B[0;32m--> 268\u001B[0m     neigh_ind \u001B[38;5;241m=\u001B[39m \u001B[38;5;28;43mself\u001B[39;49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43mkneighbors\u001B[49m\u001B[43m(\u001B[49m\u001B[43mX\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[43mreturn_distance\u001B[49m\u001B[38;5;241;43m=\u001B[39;49m\u001B[38;5;28;43;01mFalse\u001B[39;49;00m\u001B[43m)\u001B[49m\n\u001B[1;32m    269\u001B[0m     neigh_dist \u001B[38;5;241m=\u001B[39m \u001B[38;5;28;01mNone\u001B[39;00m\n\u001B[1;32m    270\u001B[0m \u001B[38;5;28;01melse\u001B[39;00m:\n",
      "File \u001B[0;32m~/opt/anaconda3/lib/python3.9/site-packages/sklearn/neighbors/_base.py:763\u001B[0m, in \u001B[0;36mKNeighborsMixin.kneighbors\u001B[0;34m(self, X, n_neighbors, return_distance)\u001B[0m\n\u001B[1;32m    756\u001B[0m use_pairwise_distances_reductions \u001B[38;5;241m=\u001B[39m (\n\u001B[1;32m    757\u001B[0m     \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39m_fit_method \u001B[38;5;241m==\u001B[39m \u001B[38;5;124m\"\u001B[39m\u001B[38;5;124mbrute\u001B[39m\u001B[38;5;124m\"\u001B[39m\n\u001B[1;32m    758\u001B[0m     \u001B[38;5;129;01mand\u001B[39;00m PairwiseDistancesArgKmin\u001B[38;5;241m.\u001B[39mis_usable_for(\n\u001B[1;32m    759\u001B[0m         X \u001B[38;5;28;01mif\u001B[39;00m X \u001B[38;5;129;01mis\u001B[39;00m \u001B[38;5;129;01mnot\u001B[39;00m \u001B[38;5;28;01mNone\u001B[39;00m \u001B[38;5;28;01melse\u001B[39;00m \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39m_fit_X, \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39m_fit_X, \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39meffective_metric_\n\u001B[1;32m    760\u001B[0m     )\n\u001B[1;32m    761\u001B[0m )\n\u001B[1;32m    762\u001B[0m \u001B[38;5;28;01mif\u001B[39;00m use_pairwise_distances_reductions:\n\u001B[0;32m--> 763\u001B[0m     results \u001B[38;5;241m=\u001B[39m \u001B[43mPairwiseDistancesArgKmin\u001B[49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43mcompute\u001B[49m\u001B[43m(\u001B[49m\n\u001B[1;32m    764\u001B[0m \u001B[43m        \u001B[49m\u001B[43mX\u001B[49m\u001B[38;5;241;43m=\u001B[39;49m\u001B[43mX\u001B[49m\u001B[43m,\u001B[49m\n\u001B[1;32m    765\u001B[0m \u001B[43m        \u001B[49m\u001B[43mY\u001B[49m\u001B[38;5;241;43m=\u001B[39;49m\u001B[38;5;28;43mself\u001B[39;49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43m_fit_X\u001B[49m\u001B[43m,\u001B[49m\n\u001B[1;32m    766\u001B[0m \u001B[43m        \u001B[49m\u001B[43mk\u001B[49m\u001B[38;5;241;43m=\u001B[39;49m\u001B[43mn_neighbors\u001B[49m\u001B[43m,\u001B[49m\n\u001B[1;32m    767\u001B[0m \u001B[43m        \u001B[49m\u001B[43mmetric\u001B[49m\u001B[38;5;241;43m=\u001B[39;49m\u001B[38;5;28;43mself\u001B[39;49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43meffective_metric_\u001B[49m\u001B[43m,\u001B[49m\n\u001B[1;32m    768\u001B[0m \u001B[43m        \u001B[49m\u001B[43mmetric_kwargs\u001B[49m\u001B[38;5;241;43m=\u001B[39;49m\u001B[38;5;28;43mself\u001B[39;49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43meffective_metric_params_\u001B[49m\u001B[43m,\u001B[49m\n\u001B[1;32m    769\u001B[0m \u001B[43m        \u001B[49m\u001B[43mstrategy\u001B[49m\u001B[38;5;241;43m=\u001B[39;49m\u001B[38;5;124;43m\"\u001B[39;49m\u001B[38;5;124;43mauto\u001B[39;49m\u001B[38;5;124;43m\"\u001B[39;49m\u001B[43m,\u001B[49m\n\u001B[1;32m    770\u001B[0m \u001B[43m        \u001B[49m\u001B[43mreturn_distance\u001B[49m\u001B[38;5;241;43m=\u001B[39;49m\u001B[43mreturn_distance\u001B[49m\u001B[43m,\u001B[49m\n\u001B[1;32m    771\u001B[0m \u001B[43m    \u001B[49m\u001B[43m)\u001B[49m\n\u001B[1;32m    773\u001B[0m \u001B[38;5;28;01melif\u001B[39;00m (\n\u001B[1;32m    774\u001B[0m     \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39m_fit_method \u001B[38;5;241m==\u001B[39m \u001B[38;5;124m\"\u001B[39m\u001B[38;5;124mbrute\u001B[39m\u001B[38;5;124m\"\u001B[39m \u001B[38;5;129;01mand\u001B[39;00m \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39mmetric \u001B[38;5;241m==\u001B[39m \u001B[38;5;124m\"\u001B[39m\u001B[38;5;124mprecomputed\u001B[39m\u001B[38;5;124m\"\u001B[39m \u001B[38;5;129;01mand\u001B[39;00m issparse(X)\n\u001B[1;32m    775\u001B[0m ):\n\u001B[1;32m    776\u001B[0m     results \u001B[38;5;241m=\u001B[39m _kneighbors_from_graph(\n\u001B[1;32m    777\u001B[0m         X, n_neighbors\u001B[38;5;241m=\u001B[39mn_neighbors, return_distance\u001B[38;5;241m=\u001B[39mreturn_distance\n\u001B[1;32m    778\u001B[0m     )\n",
      "File \u001B[0;32msklearn/metrics/_pairwise_distances_reduction.pyx:691\u001B[0m, in \u001B[0;36msklearn.metrics._pairwise_distances_reduction.PairwiseDistancesArgKmin.compute\u001B[0;34m()\u001B[0m\n",
      "File \u001B[0;32m~/opt/anaconda3/lib/python3.9/site-packages/sklearn/utils/fixes.py:151\u001B[0m, in \u001B[0;36mthreadpool_limits\u001B[0;34m(limits, user_api)\u001B[0m\n\u001B[1;32m    149\u001B[0m     \u001B[38;5;28;01mreturn\u001B[39;00m controller\u001B[38;5;241m.\u001B[39mlimit(limits\u001B[38;5;241m=\u001B[39mlimits, user_api\u001B[38;5;241m=\u001B[39muser_api)\n\u001B[1;32m    150\u001B[0m \u001B[38;5;28;01melse\u001B[39;00m:\n\u001B[0;32m--> 151\u001B[0m     \u001B[38;5;28;01mreturn\u001B[39;00m \u001B[43mthreadpoolctl\u001B[49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43mthreadpool_limits\u001B[49m\u001B[43m(\u001B[49m\u001B[43mlimits\u001B[49m\u001B[38;5;241;43m=\u001B[39;49m\u001B[43mlimits\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[43muser_api\u001B[49m\u001B[38;5;241;43m=\u001B[39;49m\u001B[43muser_api\u001B[49m\u001B[43m)\u001B[49m\n",
      "File \u001B[0;32m~/opt/anaconda3/lib/python3.9/site-packages/threadpoolctl.py:171\u001B[0m, in \u001B[0;36mthreadpool_limits.__init__\u001B[0;34m(self, limits, user_api)\u001B[0m\n\u001B[1;32m    167\u001B[0m \u001B[38;5;28;01mdef\u001B[39;00m \u001B[38;5;21m__init__\u001B[39m(\u001B[38;5;28mself\u001B[39m, limits\u001B[38;5;241m=\u001B[39m\u001B[38;5;28;01mNone\u001B[39;00m, user_api\u001B[38;5;241m=\u001B[39m\u001B[38;5;28;01mNone\u001B[39;00m):\n\u001B[1;32m    168\u001B[0m     \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39m_limits, \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39m_user_api, \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39m_prefixes \u001B[38;5;241m=\u001B[39m \\\n\u001B[1;32m    169\u001B[0m         \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39m_check_params(limits, user_api)\n\u001B[0;32m--> 171\u001B[0m     \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39m_original_info \u001B[38;5;241m=\u001B[39m \u001B[38;5;28;43mself\u001B[39;49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43m_set_threadpool_limits\u001B[49m\u001B[43m(\u001B[49m\u001B[43m)\u001B[49m\n",
      "File \u001B[0;32m~/opt/anaconda3/lib/python3.9/site-packages/threadpoolctl.py:268\u001B[0m, in \u001B[0;36mthreadpool_limits._set_threadpool_limits\u001B[0;34m(self)\u001B[0m\n\u001B[1;32m    265\u001B[0m \u001B[38;5;28;01mif\u001B[39;00m \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39m_limits \u001B[38;5;129;01mis\u001B[39;00m \u001B[38;5;28;01mNone\u001B[39;00m:\n\u001B[1;32m    266\u001B[0m     \u001B[38;5;28;01mreturn\u001B[39;00m \u001B[38;5;28;01mNone\u001B[39;00m\n\u001B[0;32m--> 268\u001B[0m modules \u001B[38;5;241m=\u001B[39m \u001B[43m_ThreadpoolInfo\u001B[49m\u001B[43m(\u001B[49m\u001B[43mprefixes\u001B[49m\u001B[38;5;241;43m=\u001B[39;49m\u001B[38;5;28;43mself\u001B[39;49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43m_prefixes\u001B[49m\u001B[43m,\u001B[49m\n\u001B[1;32m    269\u001B[0m \u001B[43m                          \u001B[49m\u001B[43muser_api\u001B[49m\u001B[38;5;241;43m=\u001B[39;49m\u001B[38;5;28;43mself\u001B[39;49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43m_user_api\u001B[49m\u001B[43m)\u001B[49m\n\u001B[1;32m    270\u001B[0m \u001B[38;5;28;01mfor\u001B[39;00m module \u001B[38;5;129;01min\u001B[39;00m modules:\n\u001B[1;32m    271\u001B[0m     \u001B[38;5;66;03m# self._limits is a dict {key: num_threads} where key is either\u001B[39;00m\n\u001B[1;32m    272\u001B[0m     \u001B[38;5;66;03m# a prefix or a user_api. If a module matches both, the limit\u001B[39;00m\n\u001B[1;32m    273\u001B[0m     \u001B[38;5;66;03m# corresponding to the prefix is chosed.\u001B[39;00m\n\u001B[1;32m    274\u001B[0m     \u001B[38;5;28;01mif\u001B[39;00m module\u001B[38;5;241m.\u001B[39mprefix \u001B[38;5;129;01min\u001B[39;00m \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39m_limits:\n",
      "File \u001B[0;32m~/opt/anaconda3/lib/python3.9/site-packages/threadpoolctl.py:340\u001B[0m, in \u001B[0;36m_ThreadpoolInfo.__init__\u001B[0;34m(self, user_api, prefixes, modules)\u001B[0m\n\u001B[1;32m    337\u001B[0m     \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39muser_api \u001B[38;5;241m=\u001B[39m [] \u001B[38;5;28;01mif\u001B[39;00m user_api \u001B[38;5;129;01mis\u001B[39;00m \u001B[38;5;28;01mNone\u001B[39;00m \u001B[38;5;28;01melse\u001B[39;00m user_api\n\u001B[1;32m    339\u001B[0m     \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39mmodules \u001B[38;5;241m=\u001B[39m []\n\u001B[0;32m--> 340\u001B[0m     \u001B[38;5;28;43mself\u001B[39;49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43m_load_modules\u001B[49m\u001B[43m(\u001B[49m\u001B[43m)\u001B[49m\n\u001B[1;32m    341\u001B[0m     \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39m_warn_if_incompatible_openmp()\n\u001B[1;32m    342\u001B[0m \u001B[38;5;28;01melse\u001B[39;00m:\n",
      "File \u001B[0;32m~/opt/anaconda3/lib/python3.9/site-packages/threadpoolctl.py:371\u001B[0m, in \u001B[0;36m_ThreadpoolInfo._load_modules\u001B[0;34m(self)\u001B[0m\n\u001B[1;32m    369\u001B[0m \u001B[38;5;124;03m\"\"\"Loop through loaded libraries and store supported ones\"\"\"\u001B[39;00m\n\u001B[1;32m    370\u001B[0m \u001B[38;5;28;01mif\u001B[39;00m sys\u001B[38;5;241m.\u001B[39mplatform \u001B[38;5;241m==\u001B[39m \u001B[38;5;124m\"\u001B[39m\u001B[38;5;124mdarwin\u001B[39m\u001B[38;5;124m\"\u001B[39m:\n\u001B[0;32m--> 371\u001B[0m     \u001B[38;5;28;43mself\u001B[39;49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43m_find_modules_with_dyld\u001B[49m\u001B[43m(\u001B[49m\u001B[43m)\u001B[49m\n\u001B[1;32m    372\u001B[0m \u001B[38;5;28;01melif\u001B[39;00m sys\u001B[38;5;241m.\u001B[39mplatform \u001B[38;5;241m==\u001B[39m \u001B[38;5;124m\"\u001B[39m\u001B[38;5;124mwin32\u001B[39m\u001B[38;5;124m\"\u001B[39m:\n\u001B[1;32m    373\u001B[0m     \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39m_find_modules_with_enum_process_module_ex()\n",
      "File \u001B[0;32m~/opt/anaconda3/lib/python3.9/site-packages/threadpoolctl.py:428\u001B[0m, in \u001B[0;36m_ThreadpoolInfo._find_modules_with_dyld\u001B[0;34m(self)\u001B[0m\n\u001B[1;32m    425\u001B[0m filepath \u001B[38;5;241m=\u001B[39m filepath\u001B[38;5;241m.\u001B[39mdecode(\u001B[38;5;124m\"\u001B[39m\u001B[38;5;124mutf-8\u001B[39m\u001B[38;5;124m\"\u001B[39m)\n\u001B[1;32m    427\u001B[0m \u001B[38;5;66;03m# Store the module if it is supported and selected\u001B[39;00m\n\u001B[0;32m--> 428\u001B[0m \u001B[38;5;28;43mself\u001B[39;49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43m_make_module_from_path\u001B[49m\u001B[43m(\u001B[49m\u001B[43mfilepath\u001B[49m\u001B[43m)\u001B[49m\n",
      "File \u001B[0;32m~/opt/anaconda3/lib/python3.9/site-packages/threadpoolctl.py:515\u001B[0m, in \u001B[0;36m_ThreadpoolInfo._make_module_from_path\u001B[0;34m(self, filepath)\u001B[0m\n\u001B[1;32m    513\u001B[0m \u001B[38;5;28;01mif\u001B[39;00m prefix \u001B[38;5;129;01min\u001B[39;00m \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39mprefixes \u001B[38;5;129;01mor\u001B[39;00m user_api \u001B[38;5;129;01min\u001B[39;00m \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39muser_api:\n\u001B[1;32m    514\u001B[0m     module_class \u001B[38;5;241m=\u001B[39m \u001B[38;5;28mglobals\u001B[39m()[module_class]\n\u001B[0;32m--> 515\u001B[0m     module \u001B[38;5;241m=\u001B[39m \u001B[43mmodule_class\u001B[49m\u001B[43m(\u001B[49m\u001B[43mfilepath\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[43mprefix\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[43muser_api\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[43minternal_api\u001B[49m\u001B[43m)\u001B[49m\n\u001B[1;32m    516\u001B[0m     \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39mmodules\u001B[38;5;241m.\u001B[39mappend(module)\n",
      "File \u001B[0;32m~/opt/anaconda3/lib/python3.9/site-packages/threadpoolctl.py:608\u001B[0m, in \u001B[0;36m_Module.__init__\u001B[0;34m(self, filepath, prefix, user_api, internal_api)\u001B[0m\n\u001B[1;32m    606\u001B[0m \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39mversion \u001B[38;5;241m=\u001B[39m \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39mget_version()\n\u001B[1;32m    607\u001B[0m \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39mnum_threads \u001B[38;5;241m=\u001B[39m \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39mget_num_threads()\n\u001B[0;32m--> 608\u001B[0m \u001B[38;5;28;43mself\u001B[39;49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43m_get_extra_info\u001B[49m\u001B[43m(\u001B[49m\u001B[43m)\u001B[49m\n",
      "File \u001B[0;32m~/opt/anaconda3/lib/python3.9/site-packages/threadpoolctl.py:662\u001B[0m, in \u001B[0;36m_OpenBLASModule._get_extra_info\u001B[0;34m(self)\u001B[0m\n\u001B[1;32m    661\u001B[0m \u001B[38;5;28;01mdef\u001B[39;00m \u001B[38;5;21m_get_extra_info\u001B[39m(\u001B[38;5;28mself\u001B[39m):\n\u001B[0;32m--> 662\u001B[0m     \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39mthreading_layer \u001B[38;5;241m=\u001B[39m \u001B[38;5;28;43mself\u001B[39;49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43mget_threading_layer\u001B[49m\u001B[43m(\u001B[49m\u001B[43m)\u001B[49m\n\u001B[1;32m    663\u001B[0m     \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39marchitecture \u001B[38;5;241m=\u001B[39m \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39mget_architecture()\n",
      "File \u001B[0;32m~/opt/anaconda3/lib/python3.9/site-packages/threadpoolctl.py:667\u001B[0m, in \u001B[0;36m_OpenBLASModule.get_threading_layer\u001B[0;34m(self)\u001B[0m\n\u001B[1;32m    665\u001B[0m \u001B[38;5;28;01mdef\u001B[39;00m \u001B[38;5;21mget_threading_layer\u001B[39m(\u001B[38;5;28mself\u001B[39m):\n\u001B[1;32m    666\u001B[0m     \u001B[38;5;124;03m\"\"\"Return the threading layer of OpenBLAS\"\"\"\u001B[39;00m\n\u001B[0;32m--> 667\u001B[0m     threading_layer \u001B[38;5;241m=\u001B[39m \u001B[38;5;28;43mself\u001B[39;49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43m_dynlib\u001B[49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43mopenblas_get_parallel\u001B[49m()\n\u001B[1;32m    668\u001B[0m     \u001B[38;5;28;01mif\u001B[39;00m threading_layer \u001B[38;5;241m==\u001B[39m \u001B[38;5;241m2\u001B[39m:\n\u001B[1;32m    669\u001B[0m         \u001B[38;5;28;01mreturn\u001B[39;00m \u001B[38;5;124m\"\u001B[39m\u001B[38;5;124mopenmp\u001B[39m\u001B[38;5;124m\"\u001B[39m\n",
      "File \u001B[0;32m~/opt/anaconda3/lib/python3.9/ctypes/__init__.py:395\u001B[0m, in \u001B[0;36mCDLL.__getattr__\u001B[0;34m(self, name)\u001B[0m\n\u001B[1;32m    393\u001B[0m \u001B[38;5;28;01mif\u001B[39;00m name\u001B[38;5;241m.\u001B[39mstartswith(\u001B[38;5;124m'\u001B[39m\u001B[38;5;124m__\u001B[39m\u001B[38;5;124m'\u001B[39m) \u001B[38;5;129;01mand\u001B[39;00m name\u001B[38;5;241m.\u001B[39mendswith(\u001B[38;5;124m'\u001B[39m\u001B[38;5;124m__\u001B[39m\u001B[38;5;124m'\u001B[39m):\n\u001B[1;32m    394\u001B[0m     \u001B[38;5;28;01mraise\u001B[39;00m \u001B[38;5;167;01mAttributeError\u001B[39;00m(name)\n\u001B[0;32m--> 395\u001B[0m func \u001B[38;5;241m=\u001B[39m \u001B[38;5;28;43mself\u001B[39;49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[38;5;21;43m__getitem__\u001B[39;49m\u001B[43m(\u001B[49m\u001B[43mname\u001B[49m\u001B[43m)\u001B[49m\n\u001B[1;32m    396\u001B[0m \u001B[38;5;28msetattr\u001B[39m(\u001B[38;5;28mself\u001B[39m, name, func)\n\u001B[1;32m    397\u001B[0m \u001B[38;5;28;01mreturn\u001B[39;00m func\n",
      "File \u001B[0;32m~/opt/anaconda3/lib/python3.9/ctypes/__init__.py:400\u001B[0m, in \u001B[0;36mCDLL.__getitem__\u001B[0;34m(self, name_or_ordinal)\u001B[0m\n\u001B[1;32m    399\u001B[0m \u001B[38;5;28;01mdef\u001B[39;00m \u001B[38;5;21m__getitem__\u001B[39m(\u001B[38;5;28mself\u001B[39m, name_or_ordinal):\n\u001B[0;32m--> 400\u001B[0m     func \u001B[38;5;241m=\u001B[39m \u001B[38;5;28;43mself\u001B[39;49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43m_FuncPtr\u001B[49m\u001B[43m(\u001B[49m\u001B[43m(\u001B[49m\u001B[43mname_or_ordinal\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[38;5;28;43mself\u001B[39;49m\u001B[43m)\u001B[49m\u001B[43m)\u001B[49m\n\u001B[1;32m    401\u001B[0m     \u001B[38;5;28;01mif\u001B[39;00m \u001B[38;5;129;01mnot\u001B[39;00m \u001B[38;5;28misinstance\u001B[39m(name_or_ordinal, \u001B[38;5;28mint\u001B[39m):\n\u001B[1;32m    402\u001B[0m         func\u001B[38;5;241m.\u001B[39m\u001B[38;5;18m__name__\u001B[39m \u001B[38;5;241m=\u001B[39m name_or_ordinal\n",
      "\u001B[0;31mAttributeError\u001B[0m: dlsym(0x9ade32b0, openblas_get_parallel): symbol not found"
     ]
    }
   ],
   "source": [
    "# bagging模型 BaggingClassifier\n",
    "# 数据2：分类\n",
    "\n",
    "from sklearn.ensemble import BaggingClassifier\n",
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "# Split the data into a training set and a test set\n",
    "X_train, X_test, y_train, y_test = train_test_split(train, target, random_state=0)\n",
    "clf = BaggingClassifier(KNeighborsClassifier(), max_samples=0.5, max_features=0.5)\n",
    "%time clf = clf.fit(X_train, y_train)\n",
    "clf.score(X_test, y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "168881e1",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 37.7 ms, sys: 267 µs, total: 38 ms\n",
      "Wall time: 38.2 ms\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.93"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 随机森林模型 RandomForestClassifier (**)\n",
    "# 数据2：分类\n",
    "\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "# Split the data into a training set and a test set\n",
    "X_train, X_test, y_train, y_test = train_test_split(train, target, random_state=0)\n",
    "clf = clf = RandomForestClassifier(n_estimators=10, max_depth=None, min_samples_split=2, random_state=0)\n",
    "%time clf = clf.fit(X_train, y_train)\n",
    "clf.score(X_test, y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "bf45d6e3",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 28.2 ms, sys: 656 µs, total: 28.9 ms\n",
      "Wall time: 28.9 ms\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.924"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 随机森林模型 ExtraTreesClassifier\n",
    "# 数据2：分类\n",
    "\n",
    "from sklearn.ensemble import ExtraTreesClassifier\n",
    "# Split the data into a training set and a test set\n",
    "X_train, X_test, y_train, y_test = train_test_split(train, target, random_state=0)\n",
    "clf = ExtraTreesClassifier(n_estimators=10, max_depth=None, min_samples_split=2, random_state=0)\n",
    "%time clf = clf.fit(X_train, y_train)\n",
    "clf.score(X_test, y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "7df5f706",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 491 ms, sys: 2.24 ms, total: 493 ms\n",
      "Wall time: 501 ms\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.928"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# AdaBoost模型 AdaBoostClassifier (*)\n",
    "# 数据2：分类\n",
    "\n",
    "from sklearn.ensemble import AdaBoostClassifier\n",
    "# Split the data into a training set and a test set\n",
    "X_train, X_test, y_train, y_test = train_test_split(train, target, random_state=0)\n",
    "clf = AdaBoostClassifier(n_estimators=100)\n",
    "%time clf = clf.fit(X_train, y_train)\n",
    "clf.score(X_test, y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "574de6f8",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 374 ms, sys: 1.61 ms, total: 375 ms\n",
      "Wall time: 384 ms\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.926"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# GradientBoosting模型 GradientBoostingClassifier (*)\n",
    "# 数据2：分类\n",
    "\n",
    "from sklearn.ensemble import GradientBoostingClassifier\n",
    "# Split the data into a training set and a test set\n",
    "X_train, X_test, y_train, y_test = train_test_split(train, target, random_state=0)\n",
    "clf = GradientBoostingClassifier(n_estimators=100, learning_rate=1.0, max_depth=1, random_state=0)\n",
    "%time clf = clf.fit(X_train, y_train)\n",
    "clf.score(X_test, y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "id": "98b7f39d",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Warning] Unknown parameter: colsample_bylevel\n",
      "[LightGBM] [Warning] Unknown parameter: silent\n",
      "[LightGBM] [Warning] Unknown parameter: tree_method\n",
      "[1]\tvalid_0's multi_logloss: 0.317017\n",
      "Training until validation scores don't improve for 100 rounds\n",
      "[2]\tvalid_0's multi_logloss: 0.317358\n",
      "[3]\tvalid_0's multi_logloss: 0.317882\n",
      "[4]\tvalid_0's multi_logloss: 0.318213\n",
      "[5]\tvalid_0's multi_logloss: 0.318525\n",
      "[6]\tvalid_0's multi_logloss: 0.318657\n",
      "[7]\tvalid_0's multi_logloss: 0.319168\n",
      "[8]\tvalid_0's multi_logloss: 0.319498\n",
      "[9]\tvalid_0's multi_logloss: 0.319846\n",
      "[10]\tvalid_0's multi_logloss: 0.319889\n",
      "[11]\tvalid_0's multi_logloss: 0.320273\n",
      "[12]\tvalid_0's multi_logloss: 0.320499\n",
      "[13]\tvalid_0's multi_logloss: 0.320718\n",
      "[14]\tvalid_0's multi_logloss: 0.321266\n",
      "[15]\tvalid_0's multi_logloss: 0.321416\n",
      "[16]\tvalid_0's multi_logloss: 0.321429\n",
      "[17]\tvalid_0's multi_logloss: 0.321682\n",
      "[18]\tvalid_0's multi_logloss: 0.321813\n",
      "[19]\tvalid_0's multi_logloss: 0.321911\n",
      "[20]\tvalid_0's multi_logloss: 0.322143\n",
      "[21]\tvalid_0's multi_logloss: 0.322415\n",
      "[22]\tvalid_0's multi_logloss: 0.32279\n",
      "[23]\tvalid_0's multi_logloss: 0.322898\n",
      "[24]\tvalid_0's multi_logloss: 0.32322\n",
      "[25]\tvalid_0's multi_logloss: 0.323686\n",
      "[26]\tvalid_0's multi_logloss: 0.323997\n",
      "[27]\tvalid_0's multi_logloss: 0.324261\n",
      "[28]\tvalid_0's multi_logloss: 0.32453\n",
      "[29]\tvalid_0's multi_logloss: 0.325055\n",
      "[30]\tvalid_0's multi_logloss: 0.325465\n",
      "[31]\tvalid_0's multi_logloss: 0.325663\n",
      "[32]\tvalid_0's multi_logloss: 0.325762\n",
      "[33]\tvalid_0's multi_logloss: 0.326032\n",
      "[34]\tvalid_0's multi_logloss: 0.326238\n",
      "[35]\tvalid_0's multi_logloss: 0.326693\n",
      "[36]\tvalid_0's multi_logloss: 0.326877\n",
      "[37]\tvalid_0's multi_logloss: 0.327292\n",
      "[38]\tvalid_0's multi_logloss: 0.327092\n",
      "[39]\tvalid_0's multi_logloss: 0.327242\n",
      "[40]\tvalid_0's multi_logloss: 0.327584\n",
      "[41]\tvalid_0's multi_logloss: 0.327999\n",
      "[42]\tvalid_0's multi_logloss: 0.328083\n",
      "[43]\tvalid_0's multi_logloss: 0.328449\n",
      "[44]\tvalid_0's multi_logloss: 0.328479\n",
      "[45]\tvalid_0's multi_logloss: 0.328786\n",
      "[46]\tvalid_0's multi_logloss: 0.329054\n",
      "[47]\tvalid_0's multi_logloss: 0.329188\n",
      "[48]\tvalid_0's multi_logloss: 0.329544\n",
      "[49]\tvalid_0's multi_logloss: 0.330031\n",
      "[50]\tvalid_0's multi_logloss: 0.330054\n",
      "[51]\tvalid_0's multi_logloss: 0.330332\n",
      "[52]\tvalid_0's multi_logloss: 0.33044\n",
      "[53]\tvalid_0's multi_logloss: 0.330584\n",
      "[54]\tvalid_0's multi_logloss: 0.330949\n",
      "[55]\tvalid_0's multi_logloss: 0.330822\n",
      "[56]\tvalid_0's multi_logloss: 0.331155\n",
      "[57]\tvalid_0's multi_logloss: 0.331267\n",
      "[58]\tvalid_0's multi_logloss: 0.331484\n",
      "[59]\tvalid_0's multi_logloss: 0.33182\n",
      "[60]\tvalid_0's multi_logloss: 0.332024\n",
      "[61]\tvalid_0's multi_logloss: 0.332283\n",
      "[62]\tvalid_0's multi_logloss: 0.332664\n",
      "[63]\tvalid_0's multi_logloss: 0.3331\n",
      "[64]\tvalid_0's multi_logloss: 0.333149\n",
      "[65]\tvalid_0's multi_logloss: 0.333434\n",
      "[66]\tvalid_0's multi_logloss: 0.333491\n",
      "[67]\tvalid_0's multi_logloss: 0.333806\n",
      "[68]\tvalid_0's multi_logloss: 0.334049\n",
      "[69]\tvalid_0's multi_logloss: 0.334534\n",
      "[70]\tvalid_0's multi_logloss: 0.334573\n",
      "[71]\tvalid_0's multi_logloss: 0.334724\n",
      "[72]\tvalid_0's multi_logloss: 0.33481\n",
      "[73]\tvalid_0's multi_logloss: 0.335113\n",
      "[74]\tvalid_0's multi_logloss: 0.335336\n",
      "[75]\tvalid_0's multi_logloss: 0.335406\n",
      "[76]\tvalid_0's multi_logloss: 0.33553\n",
      "[77]\tvalid_0's multi_logloss: 0.335857\n",
      "[78]\tvalid_0's multi_logloss: 0.33602\n",
      "[79]\tvalid_0's multi_logloss: 0.336299\n",
      "[80]\tvalid_0's multi_logloss: 0.336229\n",
      "[81]\tvalid_0's multi_logloss: 0.336497\n",
      "[82]\tvalid_0's multi_logloss: 0.336268\n",
      "[83]\tvalid_0's multi_logloss: 0.336521\n",
      "[84]\tvalid_0's multi_logloss: 0.336908\n",
      "[85]\tvalid_0's multi_logloss: 0.337197\n",
      "[86]\tvalid_0's multi_logloss: 0.337396\n",
      "[87]\tvalid_0's multi_logloss: 0.337318\n",
      "[88]\tvalid_0's multi_logloss: 0.337535\n",
      "[89]\tvalid_0's multi_logloss: 0.337266\n",
      "[90]\tvalid_0's multi_logloss: 0.337346\n",
      "[91]\tvalid_0's multi_logloss: 0.337475\n",
      "[92]\tvalid_0's multi_logloss: 0.337699\n",
      "[93]\tvalid_0's multi_logloss: 0.337878\n",
      "[94]\tvalid_0's multi_logloss: 0.337874\n",
      "[95]\tvalid_0's multi_logloss: 0.337984\n",
      "[96]\tvalid_0's multi_logloss: 0.33828\n",
      "[97]\tvalid_0's multi_logloss: 0.338536\n",
      "[98]\tvalid_0's multi_logloss: 0.338746\n",
      "[99]\tvalid_0's multi_logloss: 0.338966\n",
      "[100]\tvalid_0's multi_logloss: 0.338866\n",
      "[101]\tvalid_0's multi_logloss: 0.339012\n",
      "Early stopping, best iteration is:\n",
      "[1]\tvalid_0's multi_logloss: 0.317017\n",
      "score :  0.9475\n"
     ]
    }
   ],
   "source": [
    "# lightgbm模型 lightgbm (***)\n",
    "# 数据2：分类\n",
    "\n",
    "import lightgbm\n",
    "X_train, X_test, y_train, y_test = train_test_split(train, target, test_size=0.4, random_state=0)\n",
    "X_test, X_valid, y_test, y_valid = train_test_split(X_test, y_test, test_size=0.5, random_state=0)\n",
    "\n",
    "clf = lightgbm\n",
    "train_matrix = clf.Dataset(X_train, label=y_train)\n",
    "test_matrix = clf.Dataset(X_test, label=y_test)\n",
    "params = {'boosting_type': 'gbdt', #'boosting_type': 'dart',\n",
    "          'objective': 'multiclass', 'metric': 'multi_logloss', 'min_child_weight': 1.5, 'num_leaves': 2**5,\n",
    "          'lambda_l2': 10, 'subsample': 0.7, 'colsample_bytree': 0.7, 'colsample_bylevel': 0.7, 'learning_rate': 0.03,\n",
    "          'tree_method': 'exact', 'seed': 2017, \"num_class\": 2, 'silent': True, 'verbose': -1}\n",
    "# params = {'boosting_type': 'gbdt', 'objective': 'binary', 'eval_metric': 'auc', 'n_estimators': 200, 'max_depth': 5, 'num_leaves': 40, \n",
    "#     'max_bin': 400, 'min_data_in_leaf': 120, 'learning_rate': 0.1, 'lambda_l1': 1e-05, 'lambda_l2': 1e-05, 'min_split_gain': 0.0, \n",
    "#     'bagging_freq': 4, 'bagging_fraction': 0.9, 'feature_fraction': 0.6, 'seed': 1024, 'n_thread': 12 }\n",
    "# lightgbm.LGBMClassifier(params)\n",
    "num_round = 10000\n",
    "early_stopping_rounds = 100\n",
    "model = clf.train(params, train_matrix, num_round, valid_sets=test_matrix, early_stopping_rounds=early_stopping_rounds)\n",
    "pre= model.predict(X_valid,num_iteration=model.best_iteration)\n",
    "print('score : ', np.mean((pre[:,1]>0.5)==y_valid))\n",
    "\n",
    "# 树模型可视化\n",
    "ax = clf.plot_tree(clf, tree_index=1, figsize=(20, 8), show_info=['split_gain'])\n",
    "plt.show();"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "第1天:\n",
      "[('mean_5_day_nopromo', 720.3852458000183), ('mean_10_dow6', 691.1623229980469), ('mean_15_dow0', 506.5112819671631), ('mean_90_day', 475.885986328125), ('sum_20_dow0', 441.0540008544922), ('mean_15_dow1', 433.49130058288574), ('mean_21_day', 374.1997003555298), ('max_5_dow0', 311.4636001586914), ('mean_15_dow6', 293.37929916381836), ('nopromo_counts_365_2017', 284.44779205322266), ('mean_20_dow0', 265.6749954223633), ('sum_21_day', 229.8470001220703), ('sum_15_dow1', 221.3560028076172), ('nopromo_counts_90_2017', 201.47849655151367), ('mean_5_dow3', 189.92980098724365), ('mean_7_day', 179.37160110473633), ('day_2_hist_nopromo', 158.1943016052246), ('mean_5_dow6', 151.14758729934692), ('mean_5_dow0', 150.447998046875), ('mean_20_dow6', 149.14500427246094)]\n",
      "lightGbm mse0:   0.3521323912927727 \n",
      "\n",
      "第2天:\n",
      "[('mean_14_day_nopromo', 837.6064910888672), ('mean_5_dow6', 475.66400146484375), ('mean_15_dow1', 467.7904052734375), ('mean_5_dow3', 431.04259490966797), ('mean_30_day_nopromo', 385.22300720214844), ('mean_5_dow0', 352.0635070800781), ('sum_15_dow1', 277.93800354003906), ('day_7_hist', 258.7768964767456), ('mean_10_dow1', 253.433198928833), ('nopromo_counts_14_2017', 244.04139232635498), ('mean_5_dow2', 242.37459564208984), ('median_90_day', 196.50100708007812), ('mean_60_day_nopromo', 185.06099891662598), ('sum_20_dow6', 152.26499938964844), ('mean_5_day_nopromo', 148.92229461669922), ('median_5_dow3', 147.99120140075684), ('median_7_day_nopromo', 134.44839668273926), ('mean_7_day_nopromo', 104.31849670410156), ('sum_10_dow6', 99.8895034790039), ('nopromo_counts_90_2017', 83.08769989013672)]\n",
      "lightGbm mse1:   0.3802255050338434 \n",
      "\n",
      "第3天:\n",
      "[('max_5_dow2', 1630.7356910705566), ('mean_20_dow2', 1379.3408966064453), ('mean_5_day_nopromo', 979.6358051300049), ('day_5_hist', 531.1930961608887), ('mean_15_dow2', 467.89563512802124), ('mean_10_dow6', 347.5496997833252), ('mean_5_dow6', 309.21708059310913), ('mean_5_dow2', 256.93049716949463), ('mean_5_dow0', 221.60680389404297), ('item_nbr_5_0_max', 217.30370330810547), ('mean_30_day_nopromo', 178.47570037841797), ('sum_20_dow6', 174.0030059814453), ('sum_20_dow2', 151.4239959716797), ('item_nbr_7_0_max', 141.97847080230713), ('median_10_dow3', 135.44700622558594), ('max_5_dow6', 126.97716951370239), ('nopromo_counts_14_2017', 125.24209976196289), ('sum_10_dow6', 100.6500015258789), ('std_14_day_nopromo', 82.89652967453003), ('max_20_dow3', 78.48029899597168)]\n",
      "lightGbm mse2:   0.41625496002294143 \n",
      "\n",
      "第4天:\n",
      "[('mean_10_dow3', 2018.211503982544), ('mean_5_dow3', 718.5390949249268), ('mean_5_day_nopromo', 540.1765937805176), ('mean_5_dow4', 451.3376703262329), ('sum_5_dow3', 219.35050201416016), ('nopromo_counts_60_2017', 204.36729526519775), ('mean_5_dow0', 202.21429824829102), ('nopromo_counts_21_2017', 191.29810523986816), ('mean_10_dow4', 188.28279781341553), ('max_3_day_nopromo', 169.69829940795898), ('mean_15_dow3', 150.68499755859375), ('median_10_dow3', 148.7310028076172), ('mean_10_dow5', 147.64080238342285), ('sum_5_dow4', 143.90470123291016), ('mean_14_day_nopromo', 128.843994140625), ('mean_30_day', 125.6240005493164), ('median_5_day_nopromo', 119.43760108947754), ('nopromo_counts_30_2017', 104.65456819534302), ('max_10_dow3', 103.82450103759766), ('median_20_dow5', 97.02347040176392)]\n",
      "lightGbm mse3:   0.36372564081971037 \n",
      "\n",
      "第5天:\n",
      "[('mean_365_day_nopromo', 398.7806053161621), ('median_7_day', 227.08369636535645), ('mean_5_dow2', 172.88226127624512), ('mean_10_dow5', 157.8981113433838), ('mean_10_dow2', 155.7729034423828), ('mean_365_day', 144.77800369262695), ('median_10_dow3', 129.6072998046875), ('mean_10_dow4', 126.66211080551147), ('mean_20_dow5', 123.04970169067383), ('sum_20_dow4', 96.65410709381104), ('median_14_day', 95.80729675292969), ('day_6_hist_nopromo', 93.32626914978027), ('median_30_day', 88.83039855957031), ('max_5_dow5', 88.6352014541626), ('sum_150_day', 86.93150329589844), ('mean_10_dow3', 81.24390029907227), ('mean_20_dow4', 79.05219841003418), ('max_60_day_nopromo', 77.11731052398682), ('item_nbr_365_0_mean', 72.23989963531494), ('max_5_dow3', 69.92382955551147)]\n",
      "lightGbm mse4:   0.3007777704569458 \n",
      "\n",
      "第6天:\n",
      "[('mean_5_dow5', 835.5081901550293), ('mean_20_dow6', 502.36798095703125), ('mean_5_day_nopromo', 360.17599868774414), ('median_5_dow1', 347.9199905395508), ('sum_20_dow3', 303.61600494384766), ('day_1_hist', 296.1633014678955), ('max_5_dow1', 239.3842010498047), ('mean_5_dow0', 229.04079818725586), ('mean_20_dow5', 219.27801132202148), ('mean_15_dow6', 207.08280563354492), ('sum_20_dow6', 200.90899658203125), ('mean_30_day_nopromo', 196.49639892578125), ('nopromo_counts_365_2017', 192.99558639526367), ('sum_15_dow1', 163.9759979248047), ('mean_7_day_nopromo', 160.468994140625), ('median_20_dow5', 152.6939992904663), ('median_15_dow5', 148.88300704956055), ('median_365_day', 144.75370025634766), ('mean_365_day_nopromo', 140.61199951171875), ('mean_10_dow1', 134.56370162963867)]\n",
      "lightGbm mse5:   0.3930340982744688 \n",
      "\n",
      "第7天:\n",
      "[('mean_5_dow0', 1066.2893056869507), ('mean_30_day_nopromo', 900.8664093017578), ('mean_14_day_nopromo', 562.7765007019043), ('mean_5_dow5', 507.7863998413086), ('mean_7_day_nopromo', 428.6403408050537), ('skew_90_day_nopromo', 314.89549446105957), ('sum_5_dow0', 263.72300720214844), ('mean_15_dow3', 219.97480392456055), ('median_10_dow6', 204.3713812828064), ('mean_10_dow6', 178.89750289916992), ('std_5_dow0', 178.01320147514343), ('median_10_dow0', 162.98950386047363), ('mean_90_day_nopromo', 159.4070110321045), ('mean_21_day_nopromo', 137.8820037841797), ('nopromo_counts_14_2017', 136.26589965820312), ('mean_60_day_nopromo', 115.91698884963989), ('day_7_hist', 115.43529796600342), ('item_nbr_90_0_max', 106.35200214385986), ('sum_5_dow6', 101.50599670410156), ('min_5_day_nopromo', 82.96129989624023)]\n",
      "lightGbm mse6:   0.4273987126031383 \n",
      "\n",
      "第8天:\n",
      "[('mean_5_dow0', 786.7649984359741), ('sum_20_dow3', 485.25379180908203), ('nopromo_counts_90_2017', 469.13731384277344), ('mean_14_day_nopromo', 424.26630210876465), ('mean_10_dow1', 414.4314007759094), ('mean_60_day', 386.5229949951172), ('mean_20_dow0', 279.1160626411438), ('mean_10_dow3', 262.3884086608887), ('mean_20_dow3', 248.98849296569824), ('mean_15_dow3', 238.45849609375), ('nopromo_counts_60_2017', 231.86420249938965), ('median_10_dow3', 210.62349319458008), ('median_15_dow0', 208.01600646972656), ('mean_60_day_nopromo', 199.09679412841797), ('nopromo_counts_14_2017', 184.15160274505615), ('day_5_hist', 163.4815034866333), ('day_7_hist', 159.49462127685547), ('mean_5_dow6', 158.0428009033203), ('max_20_dow1', 108.25220012664795), ('median_10_dow6', 107.91971015930176)]\n",
      "lightGbm mse7:   0.4205397796004493 \n",
      "\n",
      "第9天:\n",
      "[('mean_10_dow1', 578.2169189453125), ('median_10_dow1', 493.8792037963867), ('mean_20_dow1', 415.0459976196289), ('sum_10_dow1', 320.3097972869873), ('median_5_dow3', 292.3363027572632), ('mean_20_dow2', 209.41230010986328), ('mean_15_dow2', 199.1212043762207), ('mean_30_day', 163.75770092010498), ('mean_60_day_nopromo', 152.24600219726562), ('mean_7_day', 150.6914005279541), ('min_5_dow1', 147.82865047454834), ('mean_20_dow6', 146.21400451660156), ('mean_10_dow5', 145.16832971572876), ('sum_5_dow1', 144.40499877929688), ('mean_10_dow2', 140.46660232543945), ('median_365_day', 135.89869689941406), ('median_5_day_nopromo', 125.12950134277344), ('median_10_dow2', 121.01599884033203), ('mean_21_day_nopromo', 120.24989700317383), ('mean_5_dow1', 116.26870346069336)]\n",
      "lightGbm mse8:   0.38008301395029614 \n",
      "\n",
      "第10天:\n",
      "[('mean_15_dow2', 1902.7278079986572), ('mean_20_dow2', 741.4533929824829), ('mean_10_dow2', 560.4538898468018), ('day_5_hist_nopromo', 513.6895799636841), ('mean_5_dow6', 503.4049987792969), ('median_10_dow2', 419.0491347312927), ('mean_10_dow3', 308.3760051727295), ('sum_20_dow2', 306.80280113220215), ('sum_15_dow2', 226.16929626464844), ('mean_21_day_nopromo', 210.28860092163086), ('sum_5_dow6', 166.49519729614258), ('sum_10_dow2', 144.97349548339844), ('mean_7_day_nopromo', 137.8984513282776), ('mean_5_day_nopromo', 122.30839920043945), ('mean_15_dow6', 120.94499969482422), ('max_5_dow2', 113.11664867401123), ('sum_21_day_nopromo', 83.88244915008545), ('mean_10_dow6', 78.39142084121704), ('mean_60_day_nopromo', 75.36899948120117), ('mean_5_dow0', 68.3304591178894)]\n",
      "lightGbm mse9:   0.4620288575364493 \n",
      "\n",
      "第11天:\n",
      "[('mean_15_dow3', 711.7960977554321), ('mean_30_day_nopromo', 598.053897857666), ('mean_10_dow6', 502.4497013092041), ('nopromo_counts_90_2017', 469.90199851989746), ('median_20_dow3', 399.1880006790161), ('sum_15_dow3', 306.03900146484375), ('max_5_dow3', 301.31763648986816), ('mean_20_dow3', 198.87320518493652), ('sum_20_dow3', 183.29029655456543), ('mean_150_decay_1', 172.9069061279297), ('sum_30_day', 166.1320037841797), ('max_14_day', 161.89989948272705), ('mean_15_dow4', 149.4209041595459), ('sum_15_dow4', 147.31129837036133), ('mean_5_dow6', 146.5792989730835), ('mean_150_day_nopromo', 133.83200073242188), ('mean_21_day_nopromo', 117.15040302276611), ('mean_30_decay_1', 102.34700012207031), ('sum_60_day', 100.35600280761719), ('mean_30_day', 88.18579864501953)]\n",
      "lightGbm mse10:   0.374985752392261 \n",
      "\n",
      "第12天:\n",
      "[('mean_5_dow5', 299.4409008026123), ('mean_20_dow4', 283.15611386299133), ('mean_60_day', 227.239501953125), ('mean_14_decay_1', 225.28949737548828), ('mean_10_dow4', 216.40180206298828), ('mean_5_dow3', 151.6102533340454), ('mean_20_dow5', 138.3861837387085), ('mean_21_decay_1', 110.2031021118164), ('mean_150_decay_1', 96.34760284423828), ('skew_150_day_nopromo', 94.88481903076172), ('max_5_dow3', 87.18765830993652), ('median_20_dow6', 75.83490180969238), ('sum_5_dow5', 72.71080207824707), ('median_3_day_nopromo', 63.33295965194702), ('mean_10_dow5', 58.374799728393555), ('max_10_dow1', 45.80989933013916), ('item_nbr_365_0_median', 40.26573896408081), ('mean_10_dow3', 36.98500061035156), ('item_nbr_90_0_mean', 36.7209997177124), ('max_5_dow5', 36.23592948913574)]\n",
      "lightGbm mse11:   0.31250332737117126 \n",
      "\n",
      "第13天:\n",
      "[('mean_10_dow5', 1208.058795928955), ('mean_15_dow5', 702.9712982177734), ('mean_15_dow6', 563.7337989807129), ('nopromo_counts_365_2017', 434.30480670928955), ('nopromo_counts_21_2017', 385.15313816070557), ('mean_5_dow5', 293.1894989013672), ('item_nbr_21_0_mean', 272.26140308380127), ('mean_20_dow5', 251.45479202270508), ('mean_15_dow0', 245.6586036682129), ('mean_15_dow1', 239.68009567260742), ('nopromo_counts_90_2017', 220.2689971923828), ('sum_15_dow5', 183.01699829101562), ('skew_150_day_nopromo', 179.46609497070312), ('mean_90_day', 141.5590057373047), ('mean_20_dow0', 137.6179962158203), ('nopromo_counts_150_2017', 135.9854326248169), ('sum_15_dow6', 119.40299987792969), ('skew_150_day', 113.16720008850098), ('mean_5_dow3', 101.50299835205078), ('day_7_hist_nopromo', 99.17010116577148)]\n",
      "lightGbm mse12:   0.4103588767130692 \n",
      "\n",
      "第14天:\n",
      "[('mean_15_dow6', 1736.7698860168457), ('mean_21_day_nopromo', 712.811393737793), ('median_10_dow6', 692.3219985961914), ('mean_20_dow6', 517.6533966064453), ('mean_15_dow5', 333.9789009094238), ('mean_10_dow3', 310.1634063720703), ('max_30_day_nopromo', 268.71730041503906), ('mean_7_day', 249.91050338745117), ('day_1_hist_nopromo', 236.6598801612854), ('max_14_day', 177.8447036743164), ('mean_3_day_nopromo', 159.67730140686035), ('mean_30_day', 154.16769981384277), ('median_150_day', 138.03599548339844), ('max_5_dow1', 134.3643980026245), ('mean_15_dow3', 128.39329624176025), ('mean_5_dow0', 126.52400207519531), ('mean_150_decay_1', 100.65110111236572), ('sum_10_dow6', 99.47916603088379), ('nopromo_counts_21_2017', 90.01902198791504), ('mean_10_dow6', 88.89588975906372)]\n",
      "lightGbm mse13:   0.45189668360942087 \n",
      "\n",
      "第15天:\n",
      "[('mean_5_dow0', 734.0249309539795), ('mean_10_dow0', 666.4844093322754), ('mean_20_dow0', 600.6020021438599), ('mean_30_day_nopromo', 491.3322067260742), ('sum_5_dow0', 334.0660037994385), ('mean_15_dow1', 310.5540008544922), ('mean_10_dow6', 280.7329921722412), ('mean_150_day_nopromo', 265.427001953125), ('mean_7_day', 217.44280242919922), ('mean_20_dow6', 208.3622055053711), ('max_10_dow0', 206.2419033050537), ('mean_15_dow6', 198.25599670410156), ('mean_21_day', 192.15080165863037), ('sum_15_dow0', 156.01800537109375), ('mean_15_dow0', 150.5658507347107), ('median_30_day_nopromo', 146.4290008544922), ('nopromo_counts_60_2017', 139.0753583908081), ('mean_10_dow5', 119.5719985961914), ('mean_7_decay_2', 114.81370162963867), ('sum_5_dow6', 102.70840072631836)]\n",
      "lightGbm mse14:   0.4334996594751623 \n",
      "\n",
      "第16天:\n",
      "[('mean_60_day_nopromo', 1610.4726028442383), ('mean_10_dow3', 593.9307060241699), ('mean_10_dow1', 481.00943660736084), ('mean_21_day', 462.7525062561035), ('nopromo_counts_30_2017', 330.9526062011719), ('mean_90_day_nopromo', 254.11740493774414), ('median_20_dow1', 207.64680099487305), ('mean_30_day', 191.3070068359375), ('max_14_day', 186.84640312194824), ('mean_5_dow0', 169.6942377090454), ('mean_5_dow5', 162.80649757385254), ('mean_10_dow0', 127.88209915161133), ('mean_21_day_nopromo', 123.85900115966797), ('sum_10_dow1', 123.49570274353027), ('median_5_dow1', 119.87030220031738), ('mean_5_dow1', 106.31070041656494), ('mean_5_dow2', 101.45635414123535), ('max_5_day_nopromo', 90.50080108642578), ('sum_10_dow5', 87.26969909667969), ('sum_20_dow3', 85.19020080566406)]\n",
      "lightGbm mse15:   0.44672814477700484 \n"
     ]
    }
   ],
   "source": [
    "# lightgbm: \n",
    "\n",
    "import lightgbm as lgb\n",
    "import sys, os\n",
    "from sklearn.metrics import mean_squared_error\n",
    "from tqdm import tqdm\n",
    "import numpy as np\n",
    "import warnings\n",
    "warnings.filterwarnings(\"ignore\")\n",
    "%matplotlib inline\n",
    "sys.stdout = sys.stdout\n",
    "\n",
    "params = {'num_leaves': 80, 'objective': 'regression', 'min_data_in_leaf': 200, 'learning_rate': 0.02, 'feature_fraction': 0.8, 'bagging_fraction': 0.7,\n",
    "    'bagging_freq': 1, 'metric': 'l2', 'n_jobs': 24, 'num_threads': -1, 'verbose': -1}\n",
    "MAX_ROUNDS = 100\n",
    "def run(i, X_train, y_train, X_val, y_val, train_weight, val_weight):\n",
    "    _original_stdout = sys.stdout # 输出重定向（设定）\n",
    "    sys.stdout = open(os.devnull, 'w') # 输出重定向（设定）\n",
    "    dtrain = lgb.Dataset(X_train, label=y_train, weight=train_weight)\n",
    "    dval = lgb.Dataset(X_val, label=y_val, reference=dtrain, weight=val_weight)\n",
    "    bst = lgb.train(params, dtrain, num_boost_round=MAX_ROUNDS, valid_sets=[dtrain, dval], verbose_eval=100)\n",
    "    val_pred = bst.predict(X_val, num_iteration=bst.best_iteration or MAX_ROUNDS)\n",
    "    score = mean_squared_error(y_val, val_pred)\n",
    "    sys.stdout.close() # 输出重定向（还原）\n",
    "    sys.stdout = _original_stdout # 输出重定向（还原）\n",
    "    print(\"第\"+str(i+1)+\"天:\")\n",
    "    print(list(sorted(zip(X_train.columns, bst.feature_importance(\"gain\")), key=lambda x: x[1], reverse=True))[:20])\n",
    "    print(\"lightGbm mse\"+str(i)+\":   \" + str(score),'\\n\\n')\n",
    "num = 1000\n",
    "for i in range(16):\n",
    "    run(i, X_train.iloc[:num,:], y_train[:num,i], X_val.iloc[:num,:], y_val[:num,i], train_weight[:num], val_weight[:num])\n"
   ],
   "metadata": {
    "collapsed": false,
    "ExecuteTime": {
     "end_time": "2023-08-26T03:25:36.620590Z",
     "start_time": "2023-08-26T03:25:29.608071Z"
    }
   },
   "id": "dfc2c935f2ce637a"
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "id": "eeaea625",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-08-25T15:34:59.256242Z",
     "start_time": "2023-08-25T15:34:27.175068Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 11.9 s, sys: 1.7 s, total: 13.6 s\n",
      "Wall time: 31.6 s\n",
      "score :  0.92\n"
     ]
    },
    {
     "data": {
      "text/plain": "<Figure size 640x480 with 1 Axes>",
      "image/png": "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\n"
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# xgboost模型 xgboost (**)\n",
    "# 数据2：分类\n",
    "\n",
    "import warnings\n",
    "warnings.filterwarnings(\"ignore\")\n",
    "%matplotlib inline\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "\n",
    "from xgboost import XGBClassifier, plot_importance #显示特征重要性\n",
    "X_train, X_test, y_train, y_test = train_test_split(train, target, test_size=0.4, random_state=0)\n",
    "clf = XGBClassifier(max_depth=2, learning_rate=1, n_estimators=1000, #弱分类树太少的话取不到更多的特征重要性\n",
    "          objective='binary:logistic')\n",
    "%time clf = clf.fit(X_train, y_train)\n",
    "print('score : ', clf.score(X_test, y_test))\n",
    "\n",
    "# 可视化特征重要性：xgboost自带的plot_importance\n",
    "plot_importance(clf); #打印重要程度结果。\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "id": "95b363ae4a00e4c0",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-08-24T06:02:58.921090600Z",
     "start_time": "2023-08-24T06:02:56.732649300Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "SGDRegressor with L2: train MSE = 0.2889191411245212 test MSE = 0.2509895595332537\n",
      "SGDRegressor with L1: train MSE = 0.288844377241948 test MSE = 0.2513945051912622\n",
      "SGDRegressor with elasticnet: train MSE = 0.2916605510761966 test MSE = 0.25271020528839755\n"
     ]
    }
   ],
   "source": [
    "# 模型正则化\n",
    "# 数据1：回归\n",
    "\n",
    "from sklearn.linear_model import SGDRegressor\n",
    "poly = PolynomialFeatures(3)\n",
    "train_data_poly = poly.fit_transform(train_data)\n",
    "test_data_poly = poly.transform(test_data)\n",
    "\n",
    "# L2范数正则化\n",
    "clf = SGDRegressor(max_iter=1000, tol=1e-3, penalty= 'L2', alpha=0.0001) \n",
    "clf.fit(train_data_poly, train_target)\n",
    "score_train = mean_squared_error(train_target, clf.predict(train_data_poly))\n",
    "score_test = mean_squared_error(test_target, clf.predict(test_data_poly))\n",
    "print(\"SGDRegressor with L2: train MSE =\", score_train, \"test MSE =\", score_test)\n",
    "\n",
    "# L1范数正则化\n",
    "clf = SGDRegressor(max_iter=1000, tol=1e-3, penalty= 'L1', alpha=0.00001) \n",
    "clf.fit(train_data_poly, train_target)\n",
    "score_train = mean_squared_error(train_target, clf.predict(train_data_poly))\n",
    "score_test = mean_squared_error(test_target, clf.predict(test_data_poly))\n",
    "print(\"SGDRegressor with L1: train MSE =\", score_train, \"test MSE =\", score_test)\n",
    "\n",
    "# ElasticNet L1和L2范数加权正则化\n",
    "clf = SGDRegressor(max_iter=1000, tol=1e-3, penalty= 'elasticnet', l1_ratio=0.9, alpha=0.00001) \n",
    "clf.fit(train_data_poly, train_target)\n",
    "score_train = mean_squared_error(train_target, clf.predict(train_data_poly))\n",
    "score_test = mean_squared_error(test_target, clf.predict(test_data_poly))\n",
    "print(\"SGDRegressor with elasticnet: train MSE =\", score_train, \"test MSE =\", score_test)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "id": "78414235e65b97ee",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-08-24T06:15:17.560822900Z",
     "start_time": "2023-08-24T06:15:17.055751300Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "GridSearchCV:\n",
      "\tParameters:{'gamma': [0.001, 0.01, 0.1, 1, 10, 100], 'C': [0.001, 0.01, 0.1, 1, 10, 100]}\n",
      "\tTest set score:0.97\n",
      "\tBest parameters:{'C': 10, 'gamma': 0.1}\n",
      "\tBest score on train set:0.98\n",
      "RandomizedSearchCV:\n",
      "\tTest set score:0.97\n",
      "\tBest parameters:{'gamma': 0.1, 'C': 10}\n",
      "\tBest score on train set:0.98\n"
     ]
    }
   ],
   "source": [
    "# 参数搜索调优: 调参\n",
    "# 数据1：回归\n",
    "\n",
    "from sklearn.datasets import load_iris\n",
    "from sklearn.svm import SVC\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "iris = load_iris()\n",
    "\n",
    "# 网格搜索 \n",
    "print('GridSearchCV:')\n",
    "param_grid = {\"gamma\":[0.001,0.01,0.1,1,10,100], \"C\":[0.001,0.01,0.1,1,10,100]}\n",
    "print(\"\\tParameters:{}\".format(param_grid))\n",
    "grid_search = GridSearchCV(SVC(),param_grid,cv=5) #实例化一个GridSearchCV类\n",
    "X_train,X_test,y_train,y_test = train_test_split(iris.data,iris.target,random_state=10)\n",
    "grid_search.fit(X_train,y_train) #训练，找到最优的参数，同时使用最优的参数实例化一个新的SVC estimator。\n",
    "# grid_search.best_estimator_\n",
    "print(\"\\tTest set score:{:.2f}\".format(grid_search.score(X_test,y_test)))\n",
    "print(\"\\tBest parameters:{}\".format(grid_search.best_params_))\n",
    "print(\"\\tBest score on train set:{:.2f}\".format(grid_search.best_score_))\n",
    "\n",
    "# 随机参数优化\n",
    "print('RandomizedSearchCV:')\n",
    "from sklearn.model_selection import RandomizedSearchCV\n",
    "param_grid = {\"gamma\":[0.001,0.01,0.1,1,10,100], \"C\":[0.001,0.01,0.1,1,10,100]}\n",
    "rs_search = RandomizedSearchCV(SVC(), param_grid, cv=5)\n",
    "rs_search.fit(X_train,y_train)\n",
    "score_test = mean_squared_error(y_test, rs_search.predict(X_test))\n",
    "print(\"\\tTest set score:{:.2f}\".format(rs_search.score(X_test,y_test)))\n",
    "print(\"\\tBest parameters:{}\".format(rs_search.best_params_))\n",
    "print(\"\\tBest score on train set:{:.2f}\".format(rs_search.best_score_))\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "81a1d9fb4b57d4f4",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training until validation scores don't improve for 100 rounds\n",
      "[50]\tTrain's l2: 0.428458\tTest's l2: 0.466965\n",
      "[100]\tTrain's l2: 0.221807\tTest's l2: 0.263431\n",
      "[150]\tTrain's l2: 0.135301\tTest's l2: 0.183515\n",
      "[200]\tTrain's l2: 0.0958737\tTest's l2: 0.148825\n",
      "[250]\tTrain's l2: 0.075642\tTest's l2: 0.133187\n",
      "[300]\tTrain's l2: 0.0629311\tTest's l2: 0.125237\n",
      "[350]\tTrain's l2: 0.054088\tTest's l2: 0.121321\n",
      "[400]\tTrain's l2: 0.0473551\tTest's l2: 0.118307\n",
      "[450]\tTrain's l2: 0.0418952\tTest's l2: 0.116414\n",
      "[500]\tTrain's l2: 0.0375119\tTest's l2: 0.114863\n",
      "[550]\tTrain's l2: 0.0337062\tTest's l2: 0.114002\n",
      "[600]\tTrain's l2: 0.0303868\tTest's l2: 0.112781\n",
      "[650]\tTrain's l2: 0.02756\tTest's l2: 0.11207\n",
      "[700]\tTrain's l2: 0.0250506\tTest's l2: 0.111287\n",
      "[750]\tTrain's l2: 0.0228758\tTest's l2: 0.110742\n",
      "[800]\tTrain's l2: 0.0209081\tTest's l2: 0.1103\n",
      "[850]\tTrain's l2: 0.0191684\tTest's l2: 0.109946\n",
      "[900]\tTrain's l2: 0.0176066\tTest's l2: 0.109759\n",
      "[950]\tTrain's l2: 0.0162243\tTest's l2: 0.109423\n",
      "[1000]\tTrain's l2: 0.014984\tTest's l2: 0.109105\n",
      "[1050]\tTrain's l2: 0.0138415\tTest's l2: 0.108795\n",
      "[1100]\tTrain's l2: 0.0128086\tTest's l2: 0.108651\n",
      "[1150]\tTrain's l2: 0.0118895\tTest's l2: 0.10835\n",
      "[1200]\tTrain's l2: 0.0110348\tTest's l2: 0.108169\n",
      "[1250]\tTrain's l2: 0.0102351\tTest's l2: 0.107997\n",
      "[1300]\tTrain's l2: 0.00949631\tTest's l2: 0.107886\n",
      "[1350]\tTrain's l2: 0.00880985\tTest's l2: 0.107708\n",
      "[1400]\tTrain's l2: 0.00818496\tTest's l2: 0.107495\n",
      "[1450]\tTrain's l2: 0.00760212\tTest's l2: 0.107251\n",
      "[1500]\tTrain's l2: 0.00706786\tTest's l2: 0.107126\n",
      "[1550]\tTrain's l2: 0.00657032\tTest's l2: 0.107007\n",
      "[1600]\tTrain's l2: 0.00612321\tTest's l2: 0.10685\n",
      "[1650]\tTrain's l2: 0.00571558\tTest's l2: 0.106697\n",
      "[1700]\tTrain's l2: 0.00532708\tTest's l2: 0.106564\n",
      "[1750]\tTrain's l2: 0.00498736\tTest's l2: 0.106505\n",
      "[1800]\tTrain's l2: 0.0046641\tTest's l2: 0.10643\n",
      "[1850]\tTrain's l2: 0.00435927\tTest's l2: 0.106395\n",
      "[1900]\tTrain's l2: 0.00407441\tTest's l2: 0.106289\n",
      "[1950]\tTrain's l2: 0.00381589\tTest's l2: 0.106228\n",
      "[2000]\tTrain's l2: 0.00358102\tTest's l2: 0.106155\n",
      "[2050]\tTrain's l2: 0.00335863\tTest's l2: 0.106083\n",
      "[2100]\tTrain's l2: 0.00314878\tTest's l2: 0.106064\n",
      "[2150]\tTrain's l2: 0.0029488\tTest's l2: 0.106039\n",
      "[2200]\tTrain's l2: 0.0027557\tTest's l2: 0.105963\n",
      "[2250]\tTrain's l2: 0.00257956\tTest's l2: 0.105857\n",
      "[2300]\tTrain's l2: 0.00242506\tTest's l2: 0.105818\n",
      "[2350]\tTrain's l2: 0.00228008\tTest's l2: 0.105762\n",
      "[2400]\tTrain's l2: 0.00213319\tTest's l2: 0.105696\n",
      "[2450]\tTrain's l2: 0.00199799\tTest's l2: 0.105679\n",
      "[2500]\tTrain's l2: 0.00187769\tTest's l2: 0.10561\n",
      "[2550]\tTrain's l2: 0.0017642\tTest's l2: 0.105564\n",
      "[2600]\tTrain's l2: 0.00165572\tTest's l2: 0.105482\n",
      "[2650]\tTrain's l2: 0.00155773\tTest's l2: 0.105454\n",
      "[2700]\tTrain's l2: 0.0014622\tTest's l2: 0.105423\n",
      "[2750]\tTrain's l2: 0.00137505\tTest's l2: 0.105434\n",
      "Early stopping, best iteration is:\n",
      "[2686]\tTrain's l2: 0.00148912\tTest's l2: 0.105398\n",
      "第1折 训练和预测 训练MSE 预测MSE\n",
      "------\n",
      " 训练MSE\n",
      " 0.0014891170026875994 \n",
      "------\n",
      "------\n",
      " 预测MSE\n",
      " 0.10539809256060709 \n",
      "------\n",
      "\n",
      "Training until validation scores don't improve for 100 rounds\n",
      "[50]\tTrain's l2: 0.436833\tTest's l2: 0.42889\n",
      "[100]\tTrain's l2: 0.221729\tTest's l2: 0.258452\n",
      "[150]\tTrain's l2: 0.132425\tTest's l2: 0.191434\n",
      "[200]\tTrain's l2: 0.0921422\tTest's l2: 0.164395\n",
      "[250]\tTrain's l2: 0.0715837\tTest's l2: 0.153253\n",
      "[300]\tTrain's l2: 0.0591252\tTest's l2: 0.147273\n",
      "[350]\tTrain's l2: 0.0505444\tTest's l2: 0.143558\n",
      "[400]\tTrain's l2: 0.0442696\tTest's l2: 0.141913\n",
      "[450]\tTrain's l2: 0.0391752\tTest's l2: 0.140459\n",
      "[500]\tTrain's l2: 0.0348435\tTest's l2: 0.139286\n",
      "[550]\tTrain's l2: 0.0313073\tTest's l2: 0.13851\n",
      "[600]\tTrain's l2: 0.0283011\tTest's l2: 0.138002\n",
      "[650]\tTrain's l2: 0.0256303\tTest's l2: 0.136922\n",
      "[700]\tTrain's l2: 0.0233832\tTest's l2: 0.1364\n",
      "[750]\tTrain's l2: 0.0213598\tTest's l2: 0.135964\n",
      "[800]\tTrain's l2: 0.0195432\tTest's l2: 0.135562\n",
      "[850]\tTrain's l2: 0.0179503\tTest's l2: 0.134965\n",
      "[900]\tTrain's l2: 0.0165388\tTest's l2: 0.134784\n",
      "[950]\tTrain's l2: 0.0152956\tTest's l2: 0.134611\n",
      "[1000]\tTrain's l2: 0.0141473\tTest's l2: 0.134473\n",
      "[1050]\tTrain's l2: 0.0130803\tTest's l2: 0.13427\n",
      "[1100]\tTrain's l2: 0.0121275\tTest's l2: 0.134171\n",
      "[1150]\tTrain's l2: 0.0112545\tTest's l2: 0.134014\n",
      "[1200]\tTrain's l2: 0.0104587\tTest's l2: 0.133869\n",
      "[1250]\tTrain's l2: 0.00970802\tTest's l2: 0.133655\n",
      "[1300]\tTrain's l2: 0.00900488\tTest's l2: 0.133415\n",
      "[1350]\tTrain's l2: 0.00834902\tTest's l2: 0.133204\n",
      "[1400]\tTrain's l2: 0.00774199\tTest's l2: 0.132926\n",
      "[1450]\tTrain's l2: 0.00720085\tTest's l2: 0.132728\n",
      "[1500]\tTrain's l2: 0.0067149\tTest's l2: 0.132755\n",
      "Early stopping, best iteration is:\n",
      "[1441]\tTrain's l2: 0.00728983\tTest's l2: 0.132697\n",
      "第2折 训练和预测 训练MSE 预测MSE\n",
      "------\n",
      " 训练MSE\n",
      " 0.0072898318628357595 \n",
      "------\n",
      "------\n",
      " 预测MSE\n",
      " 0.13269660321638646 \n",
      "------\n",
      "\n",
      "Training until validation scores don't improve for 100 rounds\n",
      "[50]\tTrain's l2: 0.435334\tTest's l2: 0.435477\n",
      "[100]\tTrain's l2: 0.224797\tTest's l2: 0.241969\n",
      "[150]\tTrain's l2: 0.136834\tTest's l2: 0.165256\n",
      "[200]\tTrain's l2: 0.0968498\tTest's l2: 0.134493\n",
      "[250]\tTrain's l2: 0.0761729\tTest's l2: 0.121531\n",
      "[300]\tTrain's l2: 0.0636853\tTest's l2: 0.11448\n",
      "[350]\tTrain's l2: 0.0546683\tTest's l2: 0.110122\n",
      "[400]\tTrain's l2: 0.0479514\tTest's l2: 0.108073\n",
      "[450]\tTrain's l2: 0.0423758\tTest's l2: 0.107005\n",
      "[500]\tTrain's l2: 0.0377975\tTest's l2: 0.105778\n",
      "[550]\tTrain's l2: 0.0339731\tTest's l2: 0.104464\n",
      "[600]\tTrain's l2: 0.0306455\tTest's l2: 0.103408\n",
      "[650]\tTrain's l2: 0.0278085\tTest's l2: 0.10251\n",
      "[700]\tTrain's l2: 0.0252935\tTest's l2: 0.101614\n",
      "[750]\tTrain's l2: 0.0231005\tTest's l2: 0.101372\n",
      "[800]\tTrain's l2: 0.021199\tTest's l2: 0.101199\n",
      "[850]\tTrain's l2: 0.0194767\tTest's l2: 0.101142\n",
      "[900]\tTrain's l2: 0.0179671\tTest's l2: 0.101309\n",
      "Early stopping, best iteration is:\n",
      "[821]\tTrain's l2: 0.020451\tTest's l2: 0.101119\n",
      "第3折 训练和预测 训练MSE 预测MSE\n",
      "------\n",
      " 训练MSE\n",
      " 0.0204510246350711 \n",
      "------\n",
      "------\n",
      " 预测MSE\n",
      " 0.10111932554955037 \n",
      "------\n",
      "\n",
      "Training until validation scores don't improve for 100 rounds\n",
      "[50]\tTrain's l2: 0.424194\tTest's l2: 0.486322\n",
      "[100]\tTrain's l2: 0.22045\tTest's l2: 0.264404\n",
      "[150]\tTrain's l2: 0.13572\tTest's l2: 0.178101\n",
      "[200]\tTrain's l2: 0.0968777\tTest's l2: 0.144088\n",
      "[250]\tTrain's l2: 0.0765346\tTest's l2: 0.129439\n",
      "[300]\tTrain's l2: 0.0639781\tTest's l2: 0.122908\n",
      "[350]\tTrain's l2: 0.0552914\tTest's l2: 0.118902\n",
      "[400]\tTrain's l2: 0.0485628\tTest's l2: 0.115741\n",
      "[450]\tTrain's l2: 0.0429733\tTest's l2: 0.114079\n",
      "[500]\tTrain's l2: 0.0383471\tTest's l2: 0.11299\n",
      "[550]\tTrain's l2: 0.0344226\tTest's l2: 0.112069\n",
      "[600]\tTrain's l2: 0.0311593\tTest's l2: 0.111212\n",
      "[650]\tTrain's l2: 0.0283902\tTest's l2: 0.110788\n",
      "[700]\tTrain's l2: 0.0258853\tTest's l2: 0.110003\n",
      "[750]\tTrain's l2: 0.0236758\tTest's l2: 0.109461\n",
      "[800]\tTrain's l2: 0.0217031\tTest's l2: 0.108945\n",
      "[850]\tTrain's l2: 0.019937\tTest's l2: 0.10856\n",
      "[900]\tTrain's l2: 0.0183338\tTest's l2: 0.108237\n",
      "[950]\tTrain's l2: 0.0168597\tTest's l2: 0.107832\n",
      "[1000]\tTrain's l2: 0.0155639\tTest's l2: 0.107673\n",
      "[1050]\tTrain's l2: 0.0143787\tTest's l2: 0.107542\n",
      "[1100]\tTrain's l2: 0.0132679\tTest's l2: 0.107417\n",
      "[1150]\tTrain's l2: 0.0123009\tTest's l2: 0.107295\n",
      "[1200]\tTrain's l2: 0.0113964\tTest's l2: 0.107125\n",
      "[1250]\tTrain's l2: 0.0105699\tTest's l2: 0.10694\n",
      "[1300]\tTrain's l2: 0.00982176\tTest's l2: 0.106933\n",
      "[1350]\tTrain's l2: 0.00913726\tTest's l2: 0.106855\n",
      "[1400]\tTrain's l2: 0.00849822\tTest's l2: 0.106897\n",
      "Early stopping, best iteration is:\n",
      "[1334]\tTrain's l2: 0.00935585\tTest's l2: 0.106822\n",
      "第4折 训练和预测 训练MSE 预测MSE\n",
      "------\n",
      " 训练MSE\n",
      " 0.009355847634266074 \n",
      "------\n",
      "------\n",
      " 预测MSE\n",
      " 0.10682190098205552 \n",
      "------\n",
      "\n",
      "Training until validation scores don't improve for 100 rounds\n",
      "[50]\tTrain's l2: 0.435061\tTest's l2: 0.446282\n",
      "[100]\tTrain's l2: 0.224844\tTest's l2: 0.252081\n",
      "[150]\tTrain's l2: 0.136923\tTest's l2: 0.172668\n",
      "[200]\tTrain's l2: 0.0962587\tTest's l2: 0.13786\n",
      "[250]\tTrain's l2: 0.0754087\tTest's l2: 0.122259\n",
      "[300]\tTrain's l2: 0.0625492\tTest's l2: 0.114111\n",
      "[350]\tTrain's l2: 0.05372\tTest's l2: 0.109744\n",
      "[400]\tTrain's l2: 0.0469531\tTest's l2: 0.106481\n",
      "[450]\tTrain's l2: 0.0415226\tTest's l2: 0.104888\n",
      "[500]\tTrain's l2: 0.0370845\tTest's l2: 0.103151\n",
      "[550]\tTrain's l2: 0.033334\tTest's l2: 0.102501\n",
      "[600]\tTrain's l2: 0.0301713\tTest's l2: 0.101926\n",
      "[650]\tTrain's l2: 0.027339\tTest's l2: 0.101613\n",
      "[700]\tTrain's l2: 0.0249321\tTest's l2: 0.10139\n",
      "[750]\tTrain's l2: 0.0228313\tTest's l2: 0.101207\n",
      "[800]\tTrain's l2: 0.0209294\tTest's l2: 0.100924\n",
      "[850]\tTrain's l2: 0.0192055\tTest's l2: 0.100718\n",
      "[900]\tTrain's l2: 0.0176946\tTest's l2: 0.100402\n",
      "[950]\tTrain's l2: 0.0163089\tTest's l2: 0.100187\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[1000]\tTrain's l2: 0.0150397\tTest's l2: 0.099986\n",
      "[1050]\tTrain's l2: 0.0138704\tTest's l2: 0.0998263\n",
      "[1100]\tTrain's l2: 0.0128394\tTest's l2: 0.0996427\n",
      "[1150]\tTrain's l2: 0.0118688\tTest's l2: 0.0996074\n",
      "[1200]\tTrain's l2: 0.0110126\tTest's l2: 0.0995783\n",
      "[1250]\tTrain's l2: 0.0102254\tTest's l2: 0.0995361\n",
      "[1300]\tTrain's l2: 0.00951201\tTest's l2: 0.0993949\n",
      "[1350]\tTrain's l2: 0.00884199\tTest's l2: 0.0992683\n",
      "[1400]\tTrain's l2: 0.00822777\tTest's l2: 0.0993064\n",
      "[1450]\tTrain's l2: 0.00767529\tTest's l2: 0.0992335\n",
      "[1500]\tTrain's l2: 0.00712999\tTest's l2: 0.0992797\n",
      "Early stopping, best iteration is:\n",
      "[1423]\tTrain's l2: 0.00797142\tTest's l2: 0.0992124\n",
      "第5折 训练和预测 训练MSE 预测MSE\n",
      "------\n",
      " 训练MSE\n",
      " 0.007971416523938925 \n",
      "------\n",
      "------\n",
      " 预测MSE\n",
      " 0.09921237825179019 \n",
      "------\n",
      "\n",
      "------\n",
      " 训练MSE\n",
      " [0.0014891170026875994, 0.0072898318628357595, 0.0204510246350711, 0.009355847634266074, 0.007971416523938925] \n",
      " 0.009311447531759892 \n",
      "------\n",
      "------\n",
      " 预测MSE\n",
      " [0.10539809256060709, 0.13269660321638646, 0.10111932554955037, 0.10682190098205552, 0.09921237825179019] \n",
      " 0.10904966011207792 \n",
      "------\n"
     ]
    }
   ],
   "source": [
    "# lightGBM 调参案例\n",
    "# 数据1：回归\n",
    "\n",
    "import pandas as pd\n",
    "train_data2 = pd.read_csv('./data/zhengqi/zhengqi_train.txt', sep='\\t')\n",
    "test_data2 = pd.read_csv('./data/zhengqi/zhengqi_test.txt', sep='\\t')\n",
    "\n",
    "train_data2_f = train_data2[test_data2.columns].values\n",
    "train_data2_target = train_data2['target'].values\n",
    "\n",
    "# lgb 模型\n",
    "from sklearn.model_selection import KFold\n",
    "from lightgbm import LGBMRegressor\n",
    "import numpy as np\n",
    "\n",
    "# 5折交叉验证\n",
    "Folds = 5\n",
    "kf = KFold(n_splits=Folds, shuffle=True, random_state=2019)\n",
    "# 记录训练和预测MSE\n",
    "MSE_DICT = {'train_mse': [], 'test_mse': []}\n",
    "\n",
    "# 线下训练预测\n",
    "for i, (train_index, test_index) in enumerate(kf.split(train_data2_f)):\n",
    "    # lgb树模型\n",
    "    lgb_reg = LGBMRegressor(learning_rate=0.01, max_depth=-1, n_estimators=5000, boosting_type='gbdt',\n",
    "        random_state=2019, objective='regression',)\n",
    "    # 切分训练集和预测集\n",
    "    X_train_KFold, X_test_KFold = train_data2_f[train_index], train_data2_f[test_index]\n",
    "    y_train_KFold, y_test_KFold = train_data2_target[train_index], train_data2_target[test_index]\n",
    "    # 训练模型\n",
    "    lgb_reg.fit(X=X_train_KFold, y=y_train_KFold,\n",
    "                eval_set=[(X_train_KFold, y_train_KFold), (X_test_KFold, y_test_KFold)],\n",
    "                eval_names=['Train', 'Test'], early_stopping_rounds=100, eval_metric='MSE', verbose=50)\n",
    "\n",
    "    # 训练集预测 测试集预测\n",
    "    y_train_KFold_predict = lgb_reg.predict(X_train_KFold, num_iteration=lgb_reg.best_iteration_)\n",
    "    y_test_KFold_predict = lgb_reg.predict(X_test_KFold, num_iteration=lgb_reg.best_iteration_)\n",
    "\n",
    "    print('第{}折 训练和预测 训练MSE 预测MSE'.format(i + 1))\n",
    "    train_mse = mean_squared_error(y_train_KFold_predict, y_train_KFold)\n",
    "    print('------\\n', '训练MSE\\n', train_mse, '\\n------')\n",
    "    test_mse = mean_squared_error(y_test_KFold_predict, y_test_KFold)\n",
    "    print('------\\n', '预测MSE\\n', test_mse, '\\n------\\n')\n",
    "\n",
    "    MSE_DICT['train_mse'].append(train_mse)\n",
    "    MSE_DICT['test_mse'].append(test_mse)\n",
    "print('------\\n', '训练MSE\\n', MSE_DICT['train_mse'], '\\n',\n",
    "      np.mean(MSE_DICT['train_mse']), '\\n------')\n",
    "print('------\\n', '预测MSE\\n', MSE_DICT['test_mse'], '\\n',\n",
    "      np.mean(MSE_DICT['test_mse']), '\\n------')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "c87f5daf",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-08-25T13:50:49.195181Z",
     "start_time": "2023-08-25T13:50:17.141954Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "rmse: mean = 0.39560381032953973, stdv = 0.000897187586141851\n",
      "rmse: mean = 0.4278982756887534, stdv = 0.003053170072809197                     \n",
      "rmse: mean = 0.37234017514299944, stdv = 0.0075128341334049775                   \n",
      "rmse: mean = 0.39583657254974436, stdv = 0.004572492833138125                    \n",
      "rmse: mean = 0.374383309403573, stdv = 0.016692697543662177                      \n",
      "rmse: mean = 0.3830427008380757, stdv = 0.0007606682887919269                    \n",
      "rmse: mean = 0.3818953412664797, stdv = 0.009277425009617896                     \n",
      "rmse: mean = 0.3609773795343162, stdv = 0.01117958110311143                      \n",
      "rmse: mean = 0.3529414388654882, stdv = 0.007893074882562523                     \n",
      "rmse: mean = 0.37825614551328457, stdv = 0.011134068090891819                   \n",
      "rmse: mean = 0.37715488353764437, stdv = 0.0026057210296076827                   \n",
      "rmse: mean = 0.3601317538172102, stdv = 0.009908512096079763                     \n",
      "rmse: mean = 0.3750499638381267, stdv = 0.014623632834307815                     \n",
      "rmse: mean = 0.3595964585684813, stdv = 0.011422976906054205                     \n",
      "rmse: mean = 0.34277677387161865, stdv = 0.010827382612207809                    \n",
      "rmse: mean = 0.35769472197760976, stdv = 0.006866819286284104                     \n",
      "rmse: mean = 0.3553933904872719, stdv = 0.007957932419063285                      \n",
      "rmse: mean = 0.34170734374844947, stdv = 0.013394014716605562                     \n",
      "rmse: mean = 0.35289382511735456, stdv = 0.01152450071027028                      \n",
      "rmse: mean = 0.3839200163309611, stdv = 0.009194061254825026                      \n",
      "rmse: mean = 0.34202261993551647, stdv = 0.011307452944544938                     \n",
      "rmse: mean = 0.3476472586540411, stdv = 0.008197223893970845                      \n",
      "rmse: mean = 0.3398449518730087, stdv = 0.011582716518462988                      \n",
      "rmse: mean = 0.33937408944899056, stdv = 0.010906213009061971                     \n",
      "rmse: mean = 0.33826627371380286, stdv = 0.011799933212644498                     \n",
      "rmse: mean = 0.3494210828861487, stdv = 0.015088662658293589                      \n",
      "rmse: mean = 0.34194079230622876, stdv = 0.011071167391022652                     \n",
      "rmse: mean = 0.36326748890541943, stdv = 0.011532554511531845                     \n",
      "rmse: mean = 0.3492581448775425, stdv = 0.010008211229249053                      \n",
      "rmse: mean = 0.33686915491584946, stdv = 0.011372926956338375                     \n",
      "100%|██████████| 30/30 [00:28<00:00,  1.04trial/s, best loss: 0.33686915491584946]\n",
      "best parameters: {'bagging_fraction': 0.8600463087344693, 'bagging_freq': 3, 'feature_fraction': 0.5556157111335985, 'learning_rate': 0.015993059191753572, 'min_child_samples': 5, 'num_leaves': 1, 'reg_alpha': 2, 'reg_lambda': 1.062178023849498}\n"
     ]
    }
   ],
   "source": [
    "# 调参案例2：lightgbm + hyperopt\n",
    "# 数据1：回归\n",
    "\n",
    "import lightgbm as lgb\n",
    "from hyperopt import hp, fmin, tpe\n",
    "from numpy.random import RandomState\n",
    "from sklearn.model_selection import KFold\n",
    "import sys, os\n",
    "\n",
    "def feature_select_wrapper(train, test, label):\n",
    "    _original_stdout = sys.stdout # 输出重定向（设定）\n",
    "    sys.stdout = open(os.devnull, 'w') # 输出重定向（设定）\n",
    "    features = train.columns.tolist()\n",
    "    features.remove(label)\n",
    "    # 配置模型的训练参数\n",
    "    params_initial = {'num_leaves': 31, 'learning_rate': 0.1, 'boosting': 'gbdt', 'min_child_samples': 20, 'bagging_seed': 2020, 'bagging_fraction': 0.7, 'bagging_freq': 1, 'feature_fraction': 0.7, 'max_depth': -1, 'metric': 'rmse', 'reg_alpha': 0, 'reg_lambda': 1, 'objective': 'regression' }\n",
    "    kf = KFold(n_splits=5, random_state=2020, shuffle=True)\n",
    "    fse = pd.Series(0, index=features)\n",
    "    for train_part_index, eval_index in kf.split(train[features], train[label]):\n",
    "        print(train[features].shape, train[label].shape)\n",
    "        # 模型训练\n",
    "        train_part = lgb.Dataset(train[features].iloc[train_part_index], train[label].iloc[train_part_index])\n",
    "        eval = lgb.Dataset(train[features].iloc[eval_index], train[label].iloc[eval_index])\n",
    "        model = lgb.train(params_initial, train_part, num_boost_round=10000, valid_sets=[train_part, eval], valid_names=['train', 'valid'], early_stopping_rounds=30, verbose_eval=50)\n",
    "        fse += pd.Series(model.feature_importance(), features)\n",
    "    feature_select = fse.sort_values(ascending=False).index.tolist()[:300]\n",
    "    sys.stdout.close() # 输出重定向（还原）\n",
    "    sys.stdout = _original_stdout # 输出重定向（还原）\n",
    "    return train[feature_select + [label]], test[feature_select], model.feature_importance()\n",
    "def params_append(params):\n",
    "    params['objective'] = 'regression'\n",
    "    params['metric'] = 'rmse'\n",
    "    params['bagging_seed'] = 2020\n",
    "    return params\n",
    "def param_hyperopt(train, label):\n",
    "    features = train.columns.tolist()\n",
    "    features.remove(label)\n",
    "    train_data = lgb.Dataset(train[features], train[label], silent=True)\n",
    "    lgb.feature_pre_filter = False\n",
    "    def hyperopt_objective(params):\n",
    "        params = params_append(params)\n",
    "        # print(params)\n",
    "        _original_stdout = sys.stdout # 输出重定向（设定）\n",
    "        sys.stdout = open(os.devnull, 'w') # 输出重定向（设定）\n",
    "        res = lgb.cv(params, train_data, 1000, nfold=2, stratified=False, shuffle=True, metrics='rmse', early_stopping_rounds=20, verbose_eval=False, show_stdv=False, seed=2020)\n",
    "        sys.stdout.close() # 输出重定向（还原）\n",
    "        sys.stdout = _original_stdout # 输出重定向（还原）\n",
    "        index = res['rmse-mean'].index(min(res['rmse-mean']))\n",
    "        print('rmse: mean = ' + str(res['rmse-mean'][index]) + ', stdv = ' + str(res['rmse-stdv'][index]))\n",
    "        return res['rmse-mean'][index]\n",
    "    params_space = {'learning_rate': hp.uniform('learning_rate', 1e-2, 5e-1), 'bagging_fraction': hp.uniform('bagging_fraction', 0.5, 1),\n",
    "        'feature_fraction': hp.uniform('feature_fraction', 0.5, 1), 'num_leaves': hp.choice('num_leaves', list(range(10, 300, 10))),\n",
    "        'reg_alpha': hp.randint('reg_alpha', 0, 10), 'reg_lambda': hp.uniform('reg_lambda', 0, 10),\n",
    "        'bagging_freq': hp.randint('bagging_freq', 1, 10), 'min_child_samples': hp.choice('min_child_samples', list(range(1, 30, 5))), 'feature_pre_filter':'false'}\n",
    "    params_best = fmin(hyperopt_objective, space=params_space, algo=tpe.suggest, max_evals=30, rstate=RandomState(2020))\n",
    "    print('best parameters:', params_best)\n",
    "    return params_best\n",
    "\n",
    "if __name__ == \"__main__\":\n",
    "    # pandas.Dataframe\n",
    "    train, test, feature_importance = feature_select_wrapper(train_data, test_data, 'target')\n",
    "    best_clf = param_hyperopt(train, 'target')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "|   iter    |  target   | colsam... |    eta    | max_depth | min_ch... | reg_alpha | reg_la... | subsample |\n",
      "-------------------------------------------------------------------------------------------------------------\n",
      "rmse: mean = 0.6939754925164031, std = 8.505373637168745e-05\n",
      "| \u001B[0m1        \u001B[0m | \u001B[0m-0.694   \u001B[0m | \u001B[0m0.5711   \u001B[0m | \u001B[0m0.1229   \u001B[0m | \u001B[0m8.665    \u001B[0m | \u001B[0m3.658    \u001B[0m | \u001B[0m4.412    \u001B[0m | \u001B[0m2.387    \u001B[0m | \u001B[0m0.9114   \u001B[0m |\n",
      "rmse: mean = 0.6973715304624193, std = 0.021168200569523454\n",
      "| \u001B[0m2        \u001B[0m | \u001B[0m-0.6974  \u001B[0m | \u001B[0m0.9934   \u001B[0m | \u001B[0m0.1842   \u001B[0m | \u001B[0m11.49    \u001B[0m | \u001B[0m27.42    \u001B[0m | \u001B[0m4.42     \u001B[0m | \u001B[0m0.5234   \u001B[0m | \u001B[0m0.5375   \u001B[0m |\n",
      "rmse: mean = 0.6692494723047557, std = 0.024692539854973883\n",
      "| \u001B[95m3        \u001B[0m | \u001B[95m-0.6692  \u001B[0m | \u001B[95m0.6078   \u001B[0m | \u001B[95m0.1822   \u001B[0m | \u001B[95m6.569    \u001B[0m | \u001B[95m6.614    \u001B[0m | \u001B[95m0.908    \u001B[0m | \u001B[95m1.192    \u001B[0m | \u001B[95m0.9216   \u001B[0m |\n",
      "rmse: mean = 0.6558967164690909, std = 0.021150634471355345\n",
      "| \u001B[95m4        \u001B[0m | \u001B[95m-0.6559  \u001B[0m | \u001B[95m0.951    \u001B[0m | \u001B[95m0.1099   \u001B[0m | \u001B[95m6.457    \u001B[0m | \u001B[95m15.72    \u001B[0m | \u001B[95m1.765    \u001B[0m | \u001B[95m1.18     \u001B[0m | \u001B[95m0.5107   \u001B[0m |\n",
      "rmse: mean = 0.6762620705919726, std = 0.021639353727133603\n",
      "| \u001B[0m5        \u001B[0m | \u001B[0m-0.6763  \u001B[0m | \u001B[0m0.7119   \u001B[0m | \u001B[0m0.0693   \u001B[0m | \u001B[0m11.66    \u001B[0m | \u001B[0m12.04    \u001B[0m | \u001B[0m2.214    \u001B[0m | \u001B[0m3.343    \u001B[0m | \u001B[0m0.9752   \u001B[0m |\n",
      "rmse: mean = 0.6614821949496803, std = 0.01674644687187732\n",
      "| \u001B[0m6        \u001B[0m | \u001B[0m-0.6615  \u001B[0m | \u001B[0m0.8542   \u001B[0m | \u001B[0m0.03426  \u001B[0m | \u001B[0m6.062    \u001B[0m | \u001B[0m14.37    \u001B[0m | \u001B[0m4.006    \u001B[0m | \u001B[0m2.779    \u001B[0m | \u001B[0m0.6056   \u001B[0m |\n",
      "rmse: mean = 0.6598950360391186, std = 0.01967633892951759\n",
      "| \u001B[0m7        \u001B[0m | \u001B[0m-0.6599  \u001B[0m | \u001B[0m0.9844   \u001B[0m | \u001B[0m0.05352  \u001B[0m | \u001B[0m9.097    \u001B[0m | \u001B[0m13.67    \u001B[0m | \u001B[0m2.489    \u001B[0m | \u001B[0m4.254    \u001B[0m | \u001B[0m0.7089   \u001B[0m |\n",
      "rmse: mean = 0.6418304938094364, std = 0.008328712150426576\n",
      "| \u001B[95m8        \u001B[0m | \u001B[95m-0.6418  \u001B[0m | \u001B[95m0.8882   \u001B[0m | \u001B[95m0.05744  \u001B[0m | \u001B[95m5.216    \u001B[0m | \u001B[95m3.579    \u001B[0m | \u001B[95m0.04138  \u001B[0m | \u001B[95m1.904    \u001B[0m | \u001B[95m0.6301   \u001B[0m |\n",
      "rmse: mean = 0.6636237505372358, std = 0.016509763738727368\n",
      "| \u001B[0m9        \u001B[0m | \u001B[0m-0.6636  \u001B[0m | \u001B[0m0.7704   \u001B[0m | \u001B[0m0.03686  \u001B[0m | \u001B[0m5.666    \u001B[0m | \u001B[0m19.71    \u001B[0m | \u001B[0m3.308    \u001B[0m | \u001B[0m3.039    \u001B[0m | \u001B[0m0.7505   \u001B[0m |\n",
      "rmse: mean = 0.6502099277209118, std = 0.020541932682625497\n",
      "| \u001B[0m10       \u001B[0m | \u001B[0m-0.6502  \u001B[0m | \u001B[0m0.502    \u001B[0m | \u001B[0m0.1503   \u001B[0m | \u001B[0m7.443    \u001B[0m | \u001B[0m10.25    \u001B[0m | \u001B[0m4.805    \u001B[0m | \u001B[0m2.423    \u001B[0m | \u001B[0m0.5675   \u001B[0m |\n",
      "rmse: mean = 0.6793046400175611, std = 0.011489846919225388\n",
      "| \u001B[0m11       \u001B[0m | \u001B[0m-0.6793  \u001B[0m | \u001B[0m0.6963   \u001B[0m | \u001B[0m0.04842  \u001B[0m | \u001B[0m10.65    \u001B[0m | \u001B[0m6.677    \u001B[0m | \u001B[0m3.779    \u001B[0m | \u001B[0m1.334    \u001B[0m | \u001B[0m0.9586   \u001B[0m |\n",
      "rmse: mean = 0.6710201010218622, std = 0.016307946139107854\n",
      "| \u001B[0m12       \u001B[0m | \u001B[0m-0.671   \u001B[0m | \u001B[0m0.53     \u001B[0m | \u001B[0m0.1571   \u001B[0m | \u001B[0m9.482    \u001B[0m | \u001B[0m4.951    \u001B[0m | \u001B[0m1.758    \u001B[0m | \u001B[0m4.642    \u001B[0m | \u001B[0m0.7558   \u001B[0m |\n",
      "rmse: mean = 0.6597891796674049, std = 0.005556110042302498\n",
      "| \u001B[0m13       \u001B[0m | \u001B[0m-0.6598  \u001B[0m | \u001B[0m0.9223   \u001B[0m | \u001B[0m0.1927   \u001B[0m | \u001B[0m6.918    \u001B[0m | \u001B[0m15.15    \u001B[0m | \u001B[0m4.254    \u001B[0m | \u001B[0m4.064    \u001B[0m | \u001B[0m0.8371   \u001B[0m |\n",
      "rmse: mean = 0.6853104249657763, std = 0.005840644748707824\n",
      "| \u001B[0m14       \u001B[0m | \u001B[0m-0.6853  \u001B[0m | \u001B[0m0.7534   \u001B[0m | \u001B[0m0.1843   \u001B[0m | \u001B[0m8.621    \u001B[0m | \u001B[0m5.037    \u001B[0m | \u001B[0m3.152    \u001B[0m | \u001B[0m4.353    \u001B[0m | \u001B[0m0.9921   \u001B[0m |\n",
      "rmse: mean = 0.6506833101251, std = 0.03262415063125684\n",
      "| \u001B[0m15       \u001B[0m | \u001B[0m-0.6507  \u001B[0m | \u001B[0m0.6797   \u001B[0m | \u001B[0m0.08165  \u001B[0m | \u001B[0m6.903    \u001B[0m | \u001B[0m20.58    \u001B[0m | \u001B[0m0.5398   \u001B[0m | \u001B[0m0.0639   \u001B[0m | \u001B[0m0.7486   \u001B[0m |\n",
      "rmse: mean = 0.6529093667152936, std = 0.014010362333626125\n",
      "| \u001B[0m16       \u001B[0m | \u001B[0m-0.6529  \u001B[0m | \u001B[0m0.7694   \u001B[0m | \u001B[0m0.1688   \u001B[0m | \u001B[0m10.21    \u001B[0m | \u001B[0m9.233    \u001B[0m | \u001B[0m0.07521  \u001B[0m | \u001B[0m4.198    \u001B[0m | \u001B[0m0.8201   \u001B[0m |\n",
      "rmse: mean = 0.6515274926377879, std = 0.018356494498306786\n",
      "| \u001B[0m17       \u001B[0m | \u001B[0m-0.6515  \u001B[0m | \u001B[0m0.8874   \u001B[0m | \u001B[0m0.02929  \u001B[0m | \u001B[0m9.551    \u001B[0m | \u001B[0m14.06    \u001B[0m | \u001B[0m1.586    \u001B[0m | \u001B[0m2.204    \u001B[0m | \u001B[0m0.7258   \u001B[0m |\n",
      "rmse: mean = 0.666595481812955, std = 0.027026554843055273\n",
      "| \u001B[0m18       \u001B[0m | \u001B[0m-0.6666  \u001B[0m | \u001B[0m0.6497   \u001B[0m | \u001B[0m0.05832  \u001B[0m | \u001B[0m5.007    \u001B[0m | \u001B[0m9.818    \u001B[0m | \u001B[0m1.685    \u001B[0m | \u001B[0m2.119    \u001B[0m | \u001B[0m0.7162   \u001B[0m |\n",
      "rmse: mean = 0.6721546782636234, std = 0.005075213890126273\n",
      "| \u001B[0m19       \u001B[0m | \u001B[0m-0.6722  \u001B[0m | \u001B[0m0.7879   \u001B[0m | \u001B[0m0.1602   \u001B[0m | \u001B[0m6.684    \u001B[0m | \u001B[0m22.68    \u001B[0m | \u001B[0m0.8951   \u001B[0m | \u001B[0m4.522    \u001B[0m | \u001B[0m0.5287   \u001B[0m |\n",
      "rmse: mean = 0.6625232086878077, std = 0.014186200653371117\n",
      "| \u001B[0m20       \u001B[0m | \u001B[0m-0.6625  \u001B[0m | \u001B[0m0.8528   \u001B[0m | \u001B[0m0.1382   \u001B[0m | \u001B[0m10.12    \u001B[0m | \u001B[0m14.93    \u001B[0m | \u001B[0m4.404    \u001B[0m | \u001B[0m3.628    \u001B[0m | \u001B[0m0.5128   \u001B[0m |\n",
      "rmse: mean = 0.6603970000591219, std = 0.01848933570387301\n",
      "| \u001B[0m21       \u001B[0m | \u001B[0m-0.6604  \u001B[0m | \u001B[0m0.6637   \u001B[0m | \u001B[0m0.06438  \u001B[0m | \u001B[0m6.804    \u001B[0m | \u001B[0m13.3     \u001B[0m | \u001B[0m2.49     \u001B[0m | \u001B[0m2.681    \u001B[0m | \u001B[0m0.8139   \u001B[0m |\n",
      "rmse: mean = 0.6701960783971609, std = 0.009532655863187967\n",
      "| \u001B[0m22       \u001B[0m | \u001B[0m-0.6702  \u001B[0m | \u001B[0m1.0      \u001B[0m | \u001B[0m0.02     \u001B[0m | \u001B[0m5.0      \u001B[0m | \u001B[0m1.062    \u001B[0m | \u001B[0m0.0      \u001B[0m | \u001B[0m1.207    \u001B[0m | \u001B[0m0.5      \u001B[0m |\n",
      "rmse: mean = 0.6596600901953877, std = 0.03215942904878849\n",
      "| \u001B[0m23       \u001B[0m | \u001B[0m-0.6597  \u001B[0m | \u001B[0m0.8236   \u001B[0m | \u001B[0m0.09246  \u001B[0m | \u001B[0m5.397    \u001B[0m | \u001B[0m4.494    \u001B[0m | \u001B[0m0.2675   \u001B[0m | \u001B[0m3.181    \u001B[0m | \u001B[0m0.5297   \u001B[0m |\n",
      "rmse: mean = 0.6810578861607515, std = 0.017458205989440034\n",
      "| \u001B[0m24       \u001B[0m | \u001B[0m-0.6811  \u001B[0m | \u001B[0m1.0      \u001B[0m | \u001B[0m0.02     \u001B[0m | \u001B[0m5.0      \u001B[0m | \u001B[0m4.116    \u001B[0m | \u001B[0m0.0      \u001B[0m | \u001B[0m0.3569   \u001B[0m | \u001B[0m1.0      \u001B[0m |\n",
      "rmse: mean = 0.648905393867286, std = 0.004456246224660787\n",
      "| \u001B[0m25       \u001B[0m | \u001B[0m-0.6489  \u001B[0m | \u001B[0m0.869    \u001B[0m | \u001B[0m0.06571  \u001B[0m | \u001B[0m5.264    \u001B[0m | \u001B[0m3.789    \u001B[0m | \u001B[0m0.09604  \u001B[0m | \u001B[0m2.245    \u001B[0m | \u001B[0m0.5981   \u001B[0m |\n",
      "rmse: mean = 0.668137424358414, std = 0.026581008413876284\n",
      "| \u001B[0m26       \u001B[0m | \u001B[0m-0.6681  \u001B[0m | \u001B[0m0.7318   \u001B[0m | \u001B[0m0.0905   \u001B[0m | \u001B[0m5.372    \u001B[0m | \u001B[0m2.887    \u001B[0m | \u001B[0m0.0      \u001B[0m | \u001B[0m2.03     \u001B[0m | \u001B[0m0.5      \u001B[0m |\n",
      "=============================================================================================================\n",
      "0.6418304938094364 {'colsample_bytree': 0.8882485288308757, 'eta': 0.0574375853997241, 'max_depth': 5.215626697787424, 'min_child_weight': 3.5791936202043386, 'reg_alpha': 0.041375907580100124, 'reg_lambda': 1.9039191474657828, 'subsample': 0.6300990594520784}\n"
     ]
    }
   ],
   "source": [
    "# 调参案例3：xgboost + BayesianOptimization\n",
    "# 数据1：回归\n",
    "\n",
    "import xgboost as xgb\n",
    "from bayes_opt import BayesianOptimization\n",
    "\n",
    "def param_beyesian(train, label):\n",
    "    features = train.columns.tolist()\n",
    "    train_y = train[label]\n",
    "    features.remove(label)\n",
    "    train_matrix = train[features].values\n",
    "    sample_index = train_y.sample(frac=0.1, random_state=2020).index.tolist()\n",
    "    train_data = xgb.DMatrix(train_matrix[sample_index, :], train_y.loc[sample_index].values, silent=True)\n",
    "    def xgb_cv(colsample_bytree, subsample, min_child_weight, max_depth, reg_alpha, eta, reg_lambda):\n",
    "        _original_stdout = sys.stdout # 输出重定向（设定）\n",
    "        sys.stdout = open(os.devnull, 'w') # 输出重定向（设定）\n",
    "        params = {'objective': 'reg:squarederror', 'early_stopping_round': 50, 'eval_metric': 'rmse'}\n",
    "        params['colsample_bytree'] = max(min(colsample_bytree, 1), 0)\n",
    "        params['subsample'] = max(min(subsample, 1), 0)\n",
    "        params[\"min_child_weight\"] = int(min_child_weight)\n",
    "        params['max_depth'] = int(max_depth)\n",
    "        params['eta'] = float(eta)\n",
    "        params['reg_alpha'] = max(reg_alpha, 0)\n",
    "        params['reg_lambda'] = max(reg_lambda, 0)\n",
    "        # print(params)\n",
    "        cv_result = xgb.cv(params, train_data, num_boost_round=1000, nfold=2, seed=2, stratified=False, shuffle=True, early_stopping_rounds=30, verbose_eval=False)\n",
    "        sys.stdout.close() # 输出重定向（还原）\n",
    "        sys.stdout = _original_stdout # 输出重定向（还原）\n",
    "        # cv_result: pandas.core.frame.DataFrame, [train-rmse-mean  train-rmse-std  test-rmse-mean  test-rmse-std]\n",
    "        test_rmse_mean, test_rmse_std = cv_result['test-rmse-mean'].tolist(), cv_result['test-rmse-std'].tolist()\n",
    "        index = test_rmse_mean.index(min(test_rmse_mean))\n",
    "        print('rmse: mean = ' + str(test_rmse_mean[index]) + ', std = ' + str(test_rmse_std[index]))\n",
    "        return -min(cv_result['test-rmse-mean'])\n",
    "    xgb_bo = BayesianOptimization(xgb_cv, {'colsample_bytree': (0.5, 1), 'subsample': (0.5, 1), 'min_child_weight': (1, 30), 'max_depth': (5, 12), 'reg_alpha': (0, 5), 'eta':(0.02, 0.2), 'reg_lambda': (0, 5)})\n",
    "    xgb_bo.maximize(init_points=21, n_iter=5)  # init_points表示初始点，n_iter代表迭代次数（即采样数）\n",
    "    print(-xgb_bo.max['target'], xgb_bo.max['params'])\n",
    "    return xgb_bo.max['params']\n",
    "\n",
    "if __name__ == \"__main__\":\n",
    "    # pandas.Dataframe\n",
    "    best_clf = param_beyesian(train, 'target')\n",
    "    "
   ],
   "metadata": {
    "collapsed": false,
    "ExecuteTime": {
     "end_time": "2023-08-25T14:23:18.979345Z",
     "start_time": "2023-08-25T14:23:11.315068Z"
    }
   },
   "id": "3a6d9949c967736c"
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Index(['V0', 'V1', 'V2', 'V3', 'V4', 'V5', 'V6', 'V7', 'V8', 'V9', 'V10',\n",
      "       'V11', 'V12', 'V13', 'V14', 'V15', 'V16', 'V17', 'V18', 'V19', 'V20',\n",
      "       'V21', 'V22', 'V23', 'V24', 'V25', 'V26', 'V27', 'V28', 'V29', 'V30',\n",
      "       'V31', 'V32', 'V33', 'V34', 'V35', 'V36', 'V37', 'target'],\n",
      "      dtype='object')\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/var/folders/6l/8jb84y794gz98hv20bv2jlmw0000gn/T/ipykernel_4617/3930985435.py:13: FutureWarning: The default dtype for empty Series will be 'object' instead of 'float64' in a future version. Specify a dtype explicitly to silence this warning.\n",
      "  prediction_train = pd.Series()\n"
     ]
    },
    {
     "ename": "CatBoostError",
     "evalue": "/Users/zomb-ml-platform-msk/go-agent-21.2.0/pipelines/BuildMaster/catboost.git/catboost/libs/train_lib/trainer_env.cpp:9: Environment for task type [GPU] not found",
     "output_type": "error",
     "traceback": [
      "\u001B[0;31m---------------------------------------------------------------------------\u001B[0m",
      "\u001B[0;31mCatBoostError\u001B[0m                             Traceback (most recent call last)",
      "Input \u001B[0;32mIn [54]\u001B[0m, in \u001B[0;36m<cell line: 27>\u001B[0;34m()\u001B[0m\n\u001B[1;32m     27\u001B[0m \u001B[38;5;28;01mif\u001B[39;00m \u001B[38;5;18m__name__\u001B[39m \u001B[38;5;241m==\u001B[39m \u001B[38;5;124m\"\u001B[39m\u001B[38;5;124m__main__\u001B[39m\u001B[38;5;124m\"\u001B[39m:\n\u001B[1;32m     28\u001B[0m     \u001B[38;5;66;03m# pandas.Dataframe\u001B[39;00m\n\u001B[1;32m     29\u001B[0m     \u001B[38;5;28mprint\u001B[39m(train_data\u001B[38;5;241m.\u001B[39mcolumns)\n\u001B[0;32m---> 30\u001B[0m     train_auc_mean, test_auc_mean \u001B[38;5;241m=\u001B[39m \u001B[43mtrain_predict\u001B[49m\u001B[43m(\u001B[49m\u001B[43mtrain_data\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[38;5;124;43m'\u001B[39;49m\u001B[38;5;124;43mtarget\u001B[39;49m\u001B[38;5;124;43m'\u001B[39;49m\u001B[43m)\u001B[49m\n",
      "Input \u001B[0;32mIn [54]\u001B[0m, in \u001B[0;36mtrain_predict\u001B[0;34m(train, label)\u001B[0m\n\u001B[1;32m     18\u001B[0m test_y \u001B[38;5;241m=\u001B[39m test_pool\u001B[38;5;241m.\u001B[39mget_label()\n\u001B[1;32m     19\u001B[0m model \u001B[38;5;241m=\u001B[39m CatBoostClassifier(iterations\u001B[38;5;241m=\u001B[39m\u001B[38;5;241m100\u001B[39m, learning_rate\u001B[38;5;241m=\u001B[39m\u001B[38;5;241m0.1\u001B[39m, eval_metric\u001B[38;5;241m=\u001B[39m\u001B[38;5;124m'\u001B[39m\u001B[38;5;124mAUC\u001B[39m\u001B[38;5;124m'\u001B[39m, task_type\u001B[38;5;241m=\u001B[39m\u001B[38;5;124m'\u001B[39m\u001B[38;5;124mGPU\u001B[39m\u001B[38;5;124m'\u001B[39m)\n\u001B[0;32m---> 20\u001B[0m \u001B[43mmodel\u001B[49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43mfit\u001B[49m\u001B[43m(\u001B[49m\u001B[43mlearn_pool\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[43meval_set\u001B[49m\u001B[38;5;241;43m=\u001B[39;49m\u001B[43mtest_pool\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[43mverbose\u001B[49m\u001B[38;5;241;43m=\u001B[39;49m\u001B[38;5;241;43m50\u001B[39;49m\u001B[43m)\u001B[49m\n\u001B[1;32m     21\u001B[0m score_train\u001B[38;5;241m.\u001B[39mappend(model\u001B[38;5;241m.\u001B[39mbest_score_[\u001B[38;5;124m'\u001B[39m\u001B[38;5;124mvalidation\u001B[39m\u001B[38;5;124m'\u001B[39m][\u001B[38;5;124m'\u001B[39m\u001B[38;5;124mAUC\u001B[39m\u001B[38;5;124m'\u001B[39m])\n\u001B[1;32m     22\u001B[0m \u001B[38;5;28mprint\u001B[39m(roc_auc_score(test_y, prediction_test))\n",
      "File \u001B[0;32m~/opt/anaconda3/lib/python3.9/site-packages/catboost/core.py:5131\u001B[0m, in \u001B[0;36mCatBoostClassifier.fit\u001B[0;34m(self, X, y, cat_features, text_features, embedding_features, sample_weight, baseline, use_best_model, eval_set, verbose, logging_level, plot, plot_file, column_description, verbose_eval, metric_period, silent, early_stopping_rounds, save_snapshot, snapshot_file, snapshot_interval, init_model, callbacks, log_cout, log_cerr)\u001B[0m\n\u001B[1;32m   5128\u001B[0m \u001B[38;5;28;01mif\u001B[39;00m \u001B[38;5;124m'\u001B[39m\u001B[38;5;124mloss_function\u001B[39m\u001B[38;5;124m'\u001B[39m \u001B[38;5;129;01min\u001B[39;00m params:\n\u001B[1;32m   5129\u001B[0m     CatBoostClassifier\u001B[38;5;241m.\u001B[39m_check_is_compatible_loss(params[\u001B[38;5;124m'\u001B[39m\u001B[38;5;124mloss_function\u001B[39m\u001B[38;5;124m'\u001B[39m])\n\u001B[0;32m-> 5131\u001B[0m \u001B[38;5;28;43mself\u001B[39;49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43m_fit\u001B[49m\u001B[43m(\u001B[49m\u001B[43mX\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[43my\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[43mcat_features\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[43mtext_features\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[43membedding_features\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[38;5;28;43;01mNone\u001B[39;49;00m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[43msample_weight\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[38;5;28;43;01mNone\u001B[39;49;00m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[38;5;28;43;01mNone\u001B[39;49;00m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[38;5;28;43;01mNone\u001B[39;49;00m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[38;5;28;43;01mNone\u001B[39;49;00m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[43mbaseline\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[43muse_best_model\u001B[49m\u001B[43m,\u001B[49m\n\u001B[1;32m   5132\u001B[0m \u001B[43m          \u001B[49m\u001B[43meval_set\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[43mverbose\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[43mlogging_level\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[43mplot\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[43mplot_file\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[43mcolumn_description\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[43mverbose_eval\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[43mmetric_period\u001B[49m\u001B[43m,\u001B[49m\n\u001B[1;32m   5133\u001B[0m \u001B[43m          \u001B[49m\u001B[43msilent\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[43mearly_stopping_rounds\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[43msave_snapshot\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[43msnapshot_file\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[43msnapshot_interval\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[43minit_model\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[43mcallbacks\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[43mlog_cout\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[43mlog_cerr\u001B[49m\u001B[43m)\u001B[49m\n\u001B[1;32m   5134\u001B[0m \u001B[38;5;28;01mreturn\u001B[39;00m \u001B[38;5;28mself\u001B[39m\n",
      "File \u001B[0;32m~/opt/anaconda3/lib/python3.9/site-packages/catboost/core.py:2357\u001B[0m, in \u001B[0;36mCatBoost._fit\u001B[0;34m(self, X, y, cat_features, text_features, embedding_features, pairs, sample_weight, group_id, group_weight, subgroup_id, pairs_weight, baseline, use_best_model, eval_set, verbose, logging_level, plot, plot_file, column_description, verbose_eval, metric_period, silent, early_stopping_rounds, save_snapshot, snapshot_file, snapshot_interval, init_model, callbacks, log_cout, log_cerr)\u001B[0m\n\u001B[1;32m   2353\u001B[0m allow_clear_pool \u001B[38;5;241m=\u001B[39m train_params[\u001B[38;5;124m\"\u001B[39m\u001B[38;5;124mallow_clear_pool\u001B[39m\u001B[38;5;124m\"\u001B[39m]\n\u001B[1;32m   2355\u001B[0m \u001B[38;5;28;01mwith\u001B[39;00m log_fixup(log_cout, log_cerr), \\\n\u001B[1;32m   2356\u001B[0m     plot_wrapper(plot, plot_file, \u001B[38;5;124m'\u001B[39m\u001B[38;5;124mTraining plots\u001B[39m\u001B[38;5;124m'\u001B[39m, [_get_train_dir(\u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39mget_params())]):\n\u001B[0;32m-> 2357\u001B[0m     \u001B[38;5;28;43mself\u001B[39;49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43m_train\u001B[49m\u001B[43m(\u001B[49m\n\u001B[1;32m   2358\u001B[0m \u001B[43m        \u001B[49m\u001B[43mtrain_pool\u001B[49m\u001B[43m,\u001B[49m\n\u001B[1;32m   2359\u001B[0m \u001B[43m        \u001B[49m\u001B[43mtrain_params\u001B[49m\u001B[43m[\u001B[49m\u001B[38;5;124;43m\"\u001B[39;49m\u001B[38;5;124;43meval_sets\u001B[39;49m\u001B[38;5;124;43m\"\u001B[39;49m\u001B[43m]\u001B[49m\u001B[43m,\u001B[49m\n\u001B[1;32m   2360\u001B[0m \u001B[43m        \u001B[49m\u001B[43mparams\u001B[49m\u001B[43m,\u001B[49m\n\u001B[1;32m   2361\u001B[0m \u001B[43m        \u001B[49m\u001B[43mallow_clear_pool\u001B[49m\u001B[43m,\u001B[49m\n\u001B[1;32m   2362\u001B[0m \u001B[43m        \u001B[49m\u001B[43mtrain_params\u001B[49m\u001B[43m[\u001B[49m\u001B[38;5;124;43m\"\u001B[39;49m\u001B[38;5;124;43minit_model\u001B[39;49m\u001B[38;5;124;43m\"\u001B[39;49m\u001B[43m]\u001B[49m\n\u001B[1;32m   2363\u001B[0m \u001B[43m    \u001B[49m\u001B[43m)\u001B[49m\n\u001B[1;32m   2365\u001B[0m \u001B[38;5;66;03m# Have property feature_importance possibly set\u001B[39;00m\n\u001B[1;32m   2366\u001B[0m loss \u001B[38;5;241m=\u001B[39m \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39m_object\u001B[38;5;241m.\u001B[39m_get_loss_function_name()\n",
      "File \u001B[0;32m~/opt/anaconda3/lib/python3.9/site-packages/catboost/core.py:1761\u001B[0m, in \u001B[0;36m_CatBoostBase._train\u001B[0;34m(self, train_pool, test_pool, params, allow_clear_pool, init_model)\u001B[0m\n\u001B[1;32m   1760\u001B[0m \u001B[38;5;28;01mdef\u001B[39;00m \u001B[38;5;21m_train\u001B[39m(\u001B[38;5;28mself\u001B[39m, train_pool, test_pool, params, allow_clear_pool, init_model):\n\u001B[0;32m-> 1761\u001B[0m     \u001B[38;5;28;43mself\u001B[39;49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43m_object\u001B[49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43m_train\u001B[49m\u001B[43m(\u001B[49m\u001B[43mtrain_pool\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[43mtest_pool\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[43mparams\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[43mallow_clear_pool\u001B[49m\u001B[43m,\u001B[49m\u001B[43m \u001B[49m\u001B[43minit_model\u001B[49m\u001B[38;5;241;43m.\u001B[39;49m\u001B[43m_object\u001B[49m\u001B[43m \u001B[49m\u001B[38;5;28;43;01mif\u001B[39;49;00m\u001B[43m \u001B[49m\u001B[43minit_model\u001B[49m\u001B[43m \u001B[49m\u001B[38;5;28;43;01melse\u001B[39;49;00m\u001B[43m \u001B[49m\u001B[38;5;28;43;01mNone\u001B[39;49;00m\u001B[43m)\u001B[49m\n\u001B[1;32m   1762\u001B[0m     \u001B[38;5;28mself\u001B[39m\u001B[38;5;241m.\u001B[39m_set_trained_model_attributes()\n",
      "File \u001B[0;32m_catboost.pyx:4625\u001B[0m, in \u001B[0;36m_catboost._CatBoost._train\u001B[0;34m()\u001B[0m\n",
      "File \u001B[0;32m_catboost.pyx:4674\u001B[0m, in \u001B[0;36m_catboost._CatBoost._train\u001B[0;34m()\u001B[0m\n",
      "\u001B[0;31mCatBoostError\u001B[0m: /Users/zomb-ml-platform-msk/go-agent-21.2.0/pipelines/BuildMaster/catboost.git/catboost/libs/train_lib/trainer_env.cpp:9: Environment for task type [GPU] not found"
     ]
    }
   ],
   "source": [
    "# 调参案例4：catboost\n",
    "# 还需要使用gpu才能跑通\n",
    "\n",
    "from sklearn.model_selection import KFold\n",
    "from catboost import Pool, CatBoostClassifier\n",
    "from sklearn.metrics import roc_auc_score\n",
    "\n",
    "def train_predict(train, label):\n",
    "    features = train.columns.tolist()\n",
    "    train_y = train[label]\n",
    "    features.remove(label)\n",
    "    kf = KFold(n_splits=5, random_state=2020, shuffle=True)\n",
    "    prediction_test = 0\n",
    "    prediction_train = pd.Series()\n",
    "    score_train = []\n",
    "    for train_part_index, eval_index in kf.split(train[features], train[label]):\n",
    "        learn_pool = Pool(train[features].iloc[train_part_index], train[label].iloc[train_part_index], text_features=[], cat_features= [], feature_names=features)\n",
    "        test_pool = Pool(train[features].iloc[eval_index], train[label].iloc[eval_index], text_features=[], cat_features= [], feature_names=features)\n",
    "        test_y = test_pool.get_label()\n",
    "        model = CatBoostClassifier(iterations=100, learning_rate=0.1, eval_metric='AUC', task_type='GPU')\n",
    "        model.fit(learn_pool, eval_set=test_pool, verbose=50)\n",
    "        score_train.append(model.best_score_['validation']['AUC'])\n",
    "        print(roc_auc_score(test_y, prediction_test))\n",
    "    train_auc_mean = sum(score_train)/5\n",
    "    print('mean auc on train  = ' + str(train_auc_mean))\n",
    "    return train_auc_mean, test_auc_mean\n",
    "\n",
    "if __name__ == \"__main__\":\n",
    "    # pandas.Dataframe\n",
    "    train_auc_mean, test_auc_mean = train_predict(train_data, 'target')\n",
    "    "
   ],
   "metadata": {
    "collapsed": false,
    "ExecuteTime": {
     "end_time": "2023-08-25T14:58:07.699292Z",
     "start_time": "2023-08-25T14:58:06.721997Z"
    }
   },
   "id": "aa50f8def34bc68"
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "outputs": [],
   "source": [],
   "metadata": {
    "collapsed": false
   },
   "id": "aa971887ac07030a"
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.9.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}