Datawhale anomalyl detection study/TASK5/TASK5_1 feature bagging.ipynb · ZhouMoqi/GitPythonShare

加入 Gitee

与超过 1200万开发者一起发现、参与优秀开源项目，私有仓库也完全免费：）

克隆/下载

TASK5_1 feature bagging.ipynb 51.05 KB

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 1.Import Library"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "from pyod.utils.data import evaluate_print# evaluate_print是基础函数所以是在pyod.utils里面，接下来是数据模块data里面\n",
    "from pyod.models.feature_bagging import FeatureBagging#model_selection是sklearn里面的，pyod是models模块\n",
    "from pyod.utils.example import visualize#visualize是基础函数里面的example模块里面的"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "from pyod.utils.data import generate_data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "from joblib import dump,load"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "from sklearn.model_selection import train_test_split"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.preprocessing import MinMaxScaler"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import precision_score\n",
    "from sklearn.metrics import roc_auc_score\n",
    "from utility import precision_n_scores"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 2. Data\n",
    "## 2.1 Generate Data \n",
    "不能什么都写成函数，只有第三次用到的时候写成函数"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "def generate_data_fun(n_train,n_test,n_features,contamination,random_state):\n",
    "    \n",
    "    x_train1,y_train1,x_test1,y_test1=\\\n",
    "    generate_data(\n",
    "    n_train=n_train,\n",
    "    n_test=n_test,\n",
    "    n_features=n_features,\n",
    "    contamination=contamination,\n",
    "     random_state=random_state)\n",
    "    return  x_train1,y_train1,x_test1,y_test1"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2.2 Import Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [],
   "source": [
    "def import_data2():\n",
    "    data2=pd.read_csv(\"breast-cancer-unsupervised-ad.csv\")\n",
    "    list_label=[]\n",
    "    for i in range(len(data2)):\n",
    "        if data2.iloc[i,30]==\"n\":\n",
    "            list_label.append(0)\n",
    "        else:\n",
    "            list_label.append(1)\n",
    "    scaler=MinMaxScaler()\n",
    "    data_x=scaler.fit_transform(data2.iloc[:,0:-1])\n",
    "    x_train2,x_test2,y_train2,y_test2=train_test_split(data_x,list_label,test_size=0.3,random_state=20)\n",
    "    return x_train2,x_test2,y_train2,y_test2\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "def import_data3():\n",
    "    data3=pd.read_excel(\"datahebing.xlsx\")\n",
    "    scaler=MinMaxScaler()\n",
    "    data_x=scaler.fit_transform(data3.iloc[:,0:-1])\n",
    "    x_train3,x_test3,y_train3,y_test3=train_test_split(data_x,data3.iloc[:,-1],test_size=0.3)\n",
    "    return x_train3,x_test3,y_train3,y_test3\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 3.Build Model\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3.1 Initialization"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3.2 Fit"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3.3 Prediction"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [],
   "source": [
    "def build_model(x_train,y_train,x_test,y_test,n_estimators=10):\n",
    "    clf_name=\"FeatureBagging\"\n",
    "    clf=FeatureBagging(contamination=0.027,check_estimator=False\n",
    "                      , n_estimators=n_estimators)\n",
    "    \n",
    "    clf.fit(x_train)\n",
    "    \n",
    "    y_train_pred=clf.labels_\n",
    "    y_train_scores=clf.decision_scores_\n",
    "    \n",
    "    y_test_pred=clf.predict(x_test)\n",
    "    y_test_scores=clf.decision_function(x_test)\n",
    "    return clf_name,y_train_pred,y_train_scores,y_test_pred,y_test_scores,clf"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 4. Evaluate Model\n",
    "## 4.1 Result"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "def evaluate_model(string,clf_name,y,y_scores):\n",
    "    print(\"\\nOn {} Data:\".format(string))\n",
    "    evaluate_print(clf_name,y,y_scores)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    " ## 4.2 Visualization"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Main Function"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "def iteneration(y_train,y_train_scores,y_test,y_test_scores):\n",
    "        roc_train=np.round(roc_auc_score(y_train, y_train_scores), decimals=4),\n",
    "        prn_train=np.round(precision_n_scores(y_train, y_train_scores), decimals=4)\n",
    "        roc_test=np.round(roc_auc_score(y_test, y_test_scores), decimals=4),\n",
    "        prn_test=np.round(precision_n_scores(y_test, y_test_scores), decimals=4)\n",
    "        return roc_train,prn_train,roc_test,prn_test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "if __name__==\"__main__\":\n",
    "    data_source=2\n",
    "    if data_source==1:\n",
    "        #1.导入数据\n",
    "        n_train=200\n",
    "        n_test=100\n",
    "        n_features=2\n",
    "        contamination=0.1\n",
    "        random_state=42\n",
    "        x_train1,y_train1,x_test1,y_test1=\\\n",
    "        generate_data_fun(n_train,n_test,n_features,contamination,random_state)\n",
    "\n",
    "        # 2.build model\n",
    "        clf_name1,y_train_pred1,y_train_scores1,y_test_pred1,y_test_scores1,clf1=\\\n",
    "        build_model(x_train1,y_train1,x_test1,y_test1)\n",
    "\n",
    "        # 3.evaluate model\n",
    "        # 3.1 result\n",
    "        string=\"Training\"\n",
    "        clf_name=clf_name1\n",
    "        y=y_train1\n",
    "        y_scores=y_train_scores1\n",
    "        evaluate_model(string,clf_name,y,y_scores)\n",
    "        evaluate_model(\"Test\",clf_name,y_test1,y_test_scores1)\n",
    "\n",
    "        # 3.2可视化\n",
    "        visualize(\n",
    "        clf_name1,\n",
    "        x_train1,\n",
    "        y_train1,\n",
    "        x_test1,\n",
    "        y_test1,\n",
    "        y_train_pred1,\n",
    "        y_test_pred1,\n",
    "        show_figure=True,\n",
    "        save_figure=False\n",
    "\n",
    "        )\n",
    "\n",
    "        # 4.save the model\n",
    "        name=\"test\"\n",
    "        model=clf1\n",
    "        dump(model,\"test\")\n",
    "    elif data_source==2:\n",
    "        roc_list_train_average=[]\n",
    "        prn_list_train_average=[]\n",
    "        roc_list_test_average=[]\n",
    "        prn_list_test_average=[]\n",
    "        for i in range(10):\n",
    "    # 注意：Python是elif，不是else if\n",
    "            roc_list_train=[]\n",
    "            prn_list_train=[]\n",
    "            roc_list_test=[]\n",
    "            prn_list_test=[]\n",
    "            for j in range(3):\n",
    "                #1.导入数据\n",
    "                x_train2,x_test2,y_train2,y_test2=import_data2()\n",
    "                # 2.build model\n",
    "                clf_name2,y_train_pred2,y_train_scores2,y_test_pred2,y_test_scores2,clf2=\\\n",
    "                build_model(x_train2,y_train2,x_test2,y_test2,10*(i+1))\n",
    "\n",
    "                # 3.evaluate model\n",
    "                # 3.1 result\n",
    "                \"\"\"\n",
    "                string=\"Training\"\n",
    "                clf_name=clf_name2\n",
    "                y=y_train2\n",
    "                y_scores=y_train_scores2\n",
    "                evaluate_model(string,clf_name,y,y_scores)\n",
    "                evaluate_model(\"Test\",clf_name,y_test2,y_test_scores2)\n",
    "                \"\"\"\n",
    "                roc_train,prn_train,roc_test,prn_test=iteneration(y_train2,y_train_scores2,y_test2,y_test_scores2)\n",
    "                roc_list_train.append(roc_train)\n",
    "                prn_list_train.append(prn_train)\n",
    "                roc_list_test.append(roc_test)\n",
    "                prn_list_test.append(prn_test)\n",
    "            roc_list_train_average.append(np.average(np.array(roc_list_train)))\n",
    "            prn_list_train_average.append(np.average(np.array(prn_list_train)))\n",
    "            roc_list_test_average.append(np.average(np.array(roc_list_test)))\n",
    "            prn_list_test_average.append(np.average(np.array(prn_list_test)))\n",
    "        plt.plot(roc_list_train_average)\n",
    "        plt.plot(prn_list_train_average)\n",
    "        plt.plot(roc_list_test_average)\n",
    "        plt.plot(prn_list_test_average)\n",
    "                 \n",
    "        # 3.2可视化\n",
    "        \"\"\"\n",
    "        visualize(\n",
    "        clf_name2,\n",
    "        np.array(x_train2)[:,0:2],\n",
    "        y_train2,\n",
    "        np.array(x_test2)[:,0:2],\n",
    "        y_test2,\n",
    "        y_train_pred2,\n",
    "        y_test_pred2,\n",
    "        show_figure=True,\n",
    "        save_figure=False\n",
    "\n",
    "        )\n",
    "        \"\"\"\n",
    "        # 4.save the model\n",
    "        #name=\"test2\"\n",
    "        #model=clf2\n",
    "        #dump(model,name)\n",
    "    elif data_source==3:\n",
    "    # 注意：Python是elif，不是else if\n",
    "        #1.导入数据\n",
    "        x_train3,x_test3,y_train3,y_test3=import_data3()\n",
    "        # 3.build model\n",
    "        clf_name3,y_train_pred3,y_train_scores3,y_test_pred3,y_test_scores3,clf3=\\\n",
    "        build_model(x_train3,y_train3,x_test3,y_test3)\n",
    "\n",
    "        # 3.evaluate model\n",
    "        # 3.1 result\n",
    "        string=\"Training\"\n",
    "        clf_name=clf_name3\n",
    "        y=y_train3\n",
    "        y_scores=y_train_scores3\n",
    "        evaluate_model(string,clf_name,y,y_scores)\n",
    "        evaluate_model(\"Test\",clf_name,y_test3,y_test_scores3)\n",
    "\n",
    "        # 3.3可视化\n",
    "        \n",
    "        visualize(\n",
    "        clf_name3,\n",
    "        np.array(x_train3)[:,0:2],\n",
    "        y_train3,\n",
    "        np.array(x_test3)[:,0:2],\n",
    "        y_test3,\n",
    "        y_train_pred3,\n",
    "        y_test_pred3,\n",
    "        show_figure=True,\n",
    "        save_figure=False\n",
    "\n",
    "        )\n",
    "\n",
    "        # 4.save the model\n",
    "        name=\"test3\"\n",
    "        model=clf3\n",
    "        dump(model,name)\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig=plt.figure(figsize=(10,10))\n",
    "plt.plot(np.arange(10,101,10),roc_list_train_average,\"r:*\",label=\"roc_train\")\n",
    "plt.plot(np.arange(10,101,10),prn_list_train_average,\"b:^\",label=\"precision_n_train\")\n",
    "plt.plot(np.arange(10,101,10),roc_list_test_average,\"g:o\",label=\"roc_test\")\n",
    "plt.plot(np.arange(10,101,10),prn_list_test_average,\"y:*\",label=\"precision_n_test\")\n",
    "plt.legend(loc=\"best\")\n",
    "plt.title(\"random_state=20\")\n",
    "plt.xlabel(\"n_estimator\")\n",
    "plt.savefig(\"random_state=20\",dpi=500)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.3"
  },
  "latex_envs": {
   "LaTeX_envs_menu_present": true,
   "autoclose": false,
   "autocomplete": true,
   "bibliofile": "biblio.bib",
   "cite_by": "apalike",
   "current_citInitial": 1,
   "eqLabelWithNumbers": true,
   "eqNumInitial": 1,
   "hotkeys": {
    "equation": "Ctrl-E",
    "itemize": "Ctrl-I"
   },
   "labels_anchors": false,
   "latex_user_defs": false,
   "report_style_numbering": false,
   "user_envs_cfg": false
  },
  "toc": {
   "base_numbering": 1,
   "nav_menu": {},
   "number_sections": true,
   "sideBar": true,
   "skip_h1_title": false,
   "title_cell": "Table of Contents",
   "title_sidebar": "Contents",
   "toc_cell": false,
   "toc_position": {
    "height": "calc(100% - 180px)",
    "left": "10px",
    "top": "150px",
    "width": "341.328px"
   },
   "toc_section_display": true,
   "toc_window_display": true
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}

一键复制编辑原始数据按行查看历史

提交于 2021-01-26 18:34 . jupyter notebook

Python

https://gitee.com/zhou-moqi/GitPythonShare.git

git@gitee.com:zhou-moqi/GitPythonShare.git

zhou-moqi

GitPythonShare

master

ZhouMoqi / GitPythonShare

简介

发行版

贡献者

近期动态

ZhouMoqi / GitPythonShare .gitee-modal { width: 500px !important; }

简介

发行版

贡献者

近期动态

搜索帮助

ZhouMoqi / GitPythonShare