预测房屋租出的概率.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 预测房屋出租状况"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 项目简介"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "一个房屋共享公司(如Airbnb)想要帮助房屋供应商为他们的房屋设定一个合理的价格。其中一个关键的步骤是建立一个模型来预测\n",
    "在一定条件下购买一个房间的可能性(以特定的特征和日期来描述)。我们的历史数据：[训练集](http://www.menet.umn.edu/~zwang/files/case2_training.csv)有50000个数据，[测试集](http://www.menet.umn.edu/~zwang/files/case2_testing.csv)有20000个数据。它们有以下特征："
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* ID:数据标识\n",
    "* Region:房屋所属的区域(取1到10之间的值的整数)\n",
    "* Date:租出的日期(1‐365之间的整数，这里我们只考虑1天的请求)\n",
    "* Weekday:星期几(1‐7之间的整数)\n",
    "* Apartment/Room:房间是整个公寓(1)还是一个房间(0)\n",
    "* Beds:房屋内床的数量(1‐4之间的整数)\n",
    "* Review：平均评分(1到5之间的连续变量)\n",
    "* Pic Quality：房屋图片的质量(0到1之间的连续变量)\n",
    "* Price:房屋的历史标价(连续变量)\n",
    "* Accept:这个房屋是否被接受(有人租了它，1)或没有(0)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "我们的目标：建立一个模型来预测每个测试集的购买可能性。我们会采用逻辑回归，决策树，随机森林，神经网络和SVM算法来分别建立模型，并选择相对优秀的模型来预测测试集。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 特征转换"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "我们首先读入训练集数据，并检查有无缺失值。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ID             0\n",
      "Region         0\n",
      "Date           0\n",
      "Weekday        0\n",
      "Apartment      0\n",
      "Beds           0\n",
      "Review         0\n",
      "Pic Quality    0\n",
      "Price          0\n",
      "Accept         0\n",
      "dtype: int64\n"
     ]
    }
   ],
   "source": [
    "#读入训练集数据\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "\n",
    "train=pd.read_csv('case2_training.csv')\n",
    "train=train.copy()\n",
    "\n",
    "#检查缺失值\n",
    "miss_n=train.isnull().sum()\n",
    "print(miss_n)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "以上可以看出训练集没有缺失值。接下来我们检查是否存在共线性的特征，通过绘出各个特征之间相关性的方法："
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#绘出各个特征之间相关系数\n",
    "import seaborn as sns\n",
    "corrmat=train.corr().abs()\n",
    "sns.heatmap(corrmat)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "由上图可知，不存在高共线性的特征。接下来我们深入各特征的商业内涵。首先是‘Price’"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x1eb1f31bb48>"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "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": [
    "train['Price'].hist()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "从上图得知：‘Price’是一个跨度比较大的连续变量，我们将它标准化。另外，‘ID’也和预测结果无关，因为它代表每个数据的标识，我们先将它删去。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "#标准化'Price'变量\n",
    "train['normalized_price'] = (train['Price'] - train['Price'].mean()) / (train['Price'].std())\n",
    "del train['Price']\n",
    "\n",
    "#删去‘ID’变量\n",
    "del train['ID']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "其次，对于‘Date’,是租出的日期(1‐365之间的整数），为了简便，我们可以将日期归类为月份（1-12）或者季度（1-4）。经检测，变季度时，模型的auc值更高，所以这里我们把它转换成季度归类。1,2,3,4分别对应着1,2,3,4季度："
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "#把日期（365天）归类转化为季度（1,2,3,4季度）\n",
    "def assign_label(h):\n",
    "    if h >=0 and h < 92:\n",
    "        return 1\n",
    "    elif h >= 92 and h <183:\n",
    "        return 2\n",
    "    elif h >=183 and h <275 :\n",
    "        return 3\n",
    "    elif h >= 275 and h < 366:\n",
    "        return 4   \n",
    "\n",
    "train[\"date_label\"] = train[\"Date\"].apply(assign_label)\n",
    "del train['Date']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "另外，对于‘Weekday’特征，我们将其所对应的1-7天转化成周末和平日。我们知道，在周末，周五和周六晚上酒店的入住率会比较高，而周日到周四晚上的平日，入住率会相对较低。因此，我们将周日到周四归为类别1，周五和周六归为类别2:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "#将周日到周四归为类别1，周五和周六归为类别2\n",
    "def assign_week(h):\n",
    "    if h >=0 and h < 5 or h>=7:\n",
    "        return 1\n",
    "    elif h >= 5 and h <7:\n",
    "        return 2\n",
    "\n",
    "train[\"week_label\"] = train[\"Weekday\"].apply(assign_week)\n",
    "del train['Weekday']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "最后，我们将原训练集里代表类别的特征‘Region’虚拟化处理："
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Apartment</th>\n",
       "      <th>Beds</th>\n",
       "      <th>Review</th>\n",
       "      <th>Pic Quality</th>\n",
       "      <th>Accept</th>\n",
       "      <th>normalized_price</th>\n",
       "      <th>date_label</th>\n",
       "      <th>week_label</th>\n",
       "      <th>Region_1</th>\n",
       "      <th>Region_2</th>\n",
       "      <th>Region_3</th>\n",
       "      <th>Region_4</th>\n",
       "      <th>Region_5</th>\n",
       "      <th>Region_6</th>\n",
       "      <th>Region_7</th>\n",
       "      <th>Region_8</th>\n",
       "      <th>Region_9</th>\n",
       "      <th>Region_10</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>4.526975</td>\n",
       "      <td>0.739994</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.932707</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>4.548455</td>\n",
       "      <td>0.862642</td>\n",
       "      <td>0</td>\n",
       "      <td>0.337202</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>4.965732</td>\n",
       "      <td>0.515548</td>\n",
       "      <td>0</td>\n",
       "      <td>1.225097</td>\n",
       "      <td>4</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>4.752735</td>\n",
       "      <td>0.561547</td>\n",
       "      <td>1</td>\n",
       "      <td>-1.809686</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>3.890960</td>\n",
       "      <td>0.817142</td>\n",
       "      <td>1</td>\n",
       "      <td>-1.637429</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Apartment  Beds    Review  Pic Quality  Accept  normalized_price  \\\n",
       "0          1     1  4.526975     0.739994       0         -0.932707   \n",
       "1          1     1  4.548455     0.862642       0          0.337202   \n",
       "2          1     2  4.965732     0.515548       0          1.225097   \n",
       "3          1     1  4.752735     0.561547       1         -1.809686   \n",
       "4          1     2  3.890960     0.817142       1         -1.637429   \n",
       "\n",
       "   date_label  week_label  Region_1  Region_2  Region_3  Region_4  Region_5  \\\n",
       "0           1           2         0         0         0         1         0   \n",
       "1           1           1         0         1         0         0         0   \n",
       "2           4           2         0         0         0         0         1   \n",
       "3           3           1         0         0         1         0         0   \n",
       "4           2           1         1         0         0         0         0   \n",
       "\n",
       "   Region_6  Region_7  Region_8  Region_9  Region_10  \n",
       "0         0         0         0         0          0  \n",
       "1         0         0         0         0          0  \n",
       "2         0         0         0         0          0  \n",
       "3         0         0         0         0          0  \n",
       "4         0         0         0         0          0  "
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#将原训练集里代表类别的特征‘Region’虚拟化处理\n",
    "dum=pd.get_dummies(train['Region'],prefix=\"Region\")\n",
    "train=pd.concat([train,dum],axis=1)\n",
    "del train['Region']\n",
    "\n",
    "#转换后的transform_train留作给不同的算法使用\n",
    "transform_train=train\n",
    "\n",
    "#显示特征转换后的前五行\n",
    "train.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "特征转换完之后，下面我们可以开始训练模型了。我们将依次用：逻辑回归，决策树，随机森林，神经网络和SVM算法来训练模型并评估。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 用逻辑回归算法来训练模型"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "我们将使用sklearn里的LogisticRegression来训练训练集；并且为了真实的描述模型的准确性，我们将执行K交叉验证：通过 sklearn.model_selection库的cross_val_predict() 函数来实现。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "#首先设置忽略warnings\n",
    "import warnings  \n",
    "warnings.filterwarnings('ignore')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "#逻辑回归算法来训练模型\n",
    "from sklearn.model_selection import cross_val_predict\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "\n",
    "train=transform_train\n",
    "#设置惩罚因子，我们加重惩罚错误预测为1的\n",
    "penalty = {\n",
    "    0: 1,\n",
    "    1: 3.6\n",
    "}\n",
    "\n",
    "#实例化模型对象:\n",
    "lr = LogisticRegression(class_weight=penalty)\n",
    "features =train.columns.drop(\"Accept\")\n",
    "lr.fit(train[features],train['Accept'])\n",
    "\n",
    "#使用5重交叉验证进行预测\n",
    "fitted_labels=cross_val_predict(lr,train[features],train['Accept'],cv=5)\n",
    "train['fitted_labels']=fitted_labels"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "训练模型后，我们将评估所预测的训练集结果。我们使用四个指标：auc(曲线下面积），accuracy(准确率），precision(精准率）和recall(召回率），四个指标的对比，我们将在后文给出。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "auc : 0.7175635602356261\n",
      "accuracy : 0.6716\n",
      "precision : 0.4419339358553947\n",
      "recall : 0.8176235572654632\n"
     ]
    },
    {
     "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": [
    "from sklearn.metrics import roc_curve,auc\n",
    "#计算auc\n",
    "fpr,tpr,thresholds =roc_curve(train['Accept'],train['fitted_labels'])\n",
    "auc=auc(fpr,tpr)\n",
    "print('auc : '+str(auc))\n",
    "\n",
    "#以fpr假正率为横轴，tpr真征率为纵轴绘出auc曲线\n",
    "plt.plot(fpr,tpr)\n",
    "plt.plot([0,1],[0,1],'r--')\n",
    "\n",
    "from sklearn.metrics import accuracy_score,precision_score,recall_score\n",
    "#计算accuracy\n",
    "accuracy = accuracy_score(train['Accept'],train['fitted_labels'])\n",
    "print('accuracy : '+str(accuracy))\n",
    "#计算precision\n",
    "precision =precision_score(train['Accept'],train['fitted_labels'])\n",
    "print('precision : '+str(precision))\n",
    "#计算recall\n",
    "recall=recall_score(train['Accept'],train['fitted_labels'])\n",
    "print('recall : '+str(recall))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "由上得出逻辑回归算法预测的auc为 0.7175635602356261。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 决策树算法"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "我们将使用sklearn里的DecisionTreeClassifier来训练训练集；同样，为了真实的描述模型的准确性，我们将执行K交叉验证：通过 sklearn.model_selection库的cross_val_predict() 函数来实现。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "#使用决策树算法来训练训练集\n",
    "from sklearn.tree import DecisionTreeClassifier\n",
    "from sklearn.model_selection import cross_val_predict\n",
    "\n",
    "train=transform_train\n",
    "\n",
    "#设置惩罚因子，我们加重惩罚错误预测为1的\n",
    "penalty = {\n",
    "    0: 1,\n",
    "    1: 3.6\n",
    "}\n",
    "\n",
    "features =train.columns.drop(\"Accept\")\n",
    "#实例化模型对象；最大树深设为10；节点被分割之前的最小行数设为2，即是具有2行的节点将不会被分割，而是变成叶子\n",
    "clf = DecisionTreeClassifier(class_weight=penalty,min_samples_split=2,max_depth=10)\n",
    "clf.fit(train[features], train['Accept'])\n",
    "\n",
    "#使用5重交叉验证进行预测\n",
    "fitted_labels=cross_val_predict(clf,train[features],train['Accept'],cv=5)\n",
    "train['fitted_labels']=fitted_labels"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "同样，训练模型后，我们将评估所预测的训练集结果。我们使用四个指标：auc(曲线下面积），accuracy(准确率），precision(精准率）和recall(召回率），四个指标的对比，我们将在后文给出。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "auc : 0.6947245848853494\n",
      "accuracy : 0.63256\n",
      "precision : 0.4110427200117242\n",
      "recall : 0.8300532701982836\n"
     ]
    },
    {
     "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": [
    "from sklearn.metrics import roc_curve,auc\n",
    "#计算auc\n",
    "fpr,tpr,thresholds =roc_curve(train['Accept'],train['fitted_labels'])\n",
    "auc=auc(fpr,tpr)\n",
    "print('auc : '+str(auc))\n",
    "\n",
    "#以fpr假正率为横轴，tpr真征率为纵轴绘出auc曲线\n",
    "plt.plot(fpr,tpr)\n",
    "plt.plot([0,1],[0,1],'r--')\n",
    "\n",
    "from sklearn.metrics import accuracy_score,precision_score,recall_score\n",
    "#计算accuracy\n",
    "accuracy = accuracy_score(train['Accept'],train['fitted_labels'])\n",
    "print('accuracy : '+str(accuracy))\n",
    "#计算precision\n",
    "precision =precision_score(train['Accept'],train['fitted_labels'])\n",
    "print('precision : '+str(precision))\n",
    "#计算recall\n",
    "recall=recall_score(train['Accept'],train['fitted_labels'])\n",
    "print('recall : '+str(recall))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "由上得出决策树算法预测的auc为 0.6948835773388841。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 随机森林算法"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "通常，随机森林能减少过度拟合，并且总体上提高模型的准确性。我们将使用sklearn.ensemble里的RandomForestClassifier来训练训练集；同样，为了真实的描述模型的准确性，我们将执行K交叉验证：通过 sklearn.model_selection库的cross_val_predict() 函数来实现。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "#使用随机森林算法来训练训练集\n",
    "from sklearn.model_selection import cross_val_predict\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "\n",
    "train=transform_train\n",
    "\n",
    "#设置惩罚因子，我们加重惩罚错误预测为1的\n",
    "penalty = {\n",
    "    0: 1,\n",
    "    1: 3.6\n",
    "}\n",
    "\n",
    "#实例化模型对象：最大树深设为11；节点被分割之前的最小行数设为2，即是具有2行的节点将不会被分割，而是变成叶子\n",
    "rf =RandomForestClassifier(class_weight=penalty,min_samples_split=2,max_depth=11)\n",
    "features =train.columns.drop(\"Accept\")\n",
    "rf.fit(train[features],train['Accept'])\n",
    "\n",
    "#使用5重交叉验证进行预测\n",
    "fitted_labels=cross_val_predict(rf,train[features],train['Accept'],cv=5)\n",
    "train['fitted_labels']=fitted_labels"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "同样，训练模型后，我们将评估所预测的训练集结果。我们使用四个指标：auc(曲线下面积），accuracy(准确率），precision(精准率）和recall(召回率），四个指标的对比，我们将在后文给出。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "auc : 0.7010618705296477\n",
      "accuracy : 0.64276\n",
      "precision : 0.4186994911703083\n",
      "recall : 0.8279816513761468\n"
     ]
    },
    {
     "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": [
    "from sklearn.metrics import roc_curve,auc\n",
    "#计算auc\n",
    "fpr,tpr,thresholds =roc_curve(train['Accept'],train['fitted_labels'])\n",
    "auc=auc(fpr,tpr)\n",
    "print('auc : '+str(auc))\n",
    "\n",
    "#以fpr假正率为横轴，tpr真征率为纵轴绘出auc曲线\n",
    "plt.plot(fpr,tpr)\n",
    "plt.plot([0,1],[0,1],'r--')\n",
    "\n",
    "from sklearn.metrics import accuracy_score,precision_score,recall_score\n",
    "#计算accuracy\n",
    "accuracy = accuracy_score(train['Accept'],train['fitted_labels'])\n",
    "print('accuracy : '+str(accuracy))\n",
    "#计算precision\n",
    "precision =precision_score(train['Accept'],train['fitted_labels'])\n",
    "print('precision : '+str(precision))\n",
    "#计算recall\n",
    "recall=recall_score(train['Accept'],train['fitted_labels'])\n",
    "print('recall : '+str(recall))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "由上我们得出，随机森林算法的auc为 0.7010618705296477"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 神经网络"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "我们将使用sklearn.neural_network里的MLPClassifier来训练训练集；同样，为了真实的描述模型的准确性，我们将执行K交叉验证：通过 sklearn.model_selection库的cross_val_predict() 函数来实现。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "#使用神经网络来训练训练集\n",
    "from sklearn.neural_network import MLPClassifier\n",
    "from sklearn.model_selection import cross_val_predict\n",
    "\n",
    "train=transform_train\n",
    "\n",
    "features =train.columns.drop('Accept')\n",
    "\n",
    "#实例化模型对象：隐藏层的数量是3，没层的神经元数量分别是2,6,10；\n",
    "#使用激活参数指定所有层中的激活函数:logistic：sigmoid函数；tanh：双曲正切函数；relu:ReLU函数;identity：正比例函数\n",
    "mlp=MLPClassifier(hidden_layer_sizes=(2,6,10),activation='logistic')\n",
    "mlp.fit(train[features],train['Accept'])\n",
    "\n",
    "#使用5重交叉验证进行预测\n",
    "fitted_labels=cross_val_predict(mlp,train[features],train['Accept'],cv=4)\n",
    "train['fitted_labels']=fitted_labels"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "同样，训练模型后，我们将评估所预测的训练集结果。我们使用四个指标：auc(曲线下面积），accuracy(准确率），precision(精准率）和recall(召回率），四个指标的对比，我们将在后文给出。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "神经网络：\n",
      "auc : 0.5878284720575782\n",
      "accuracy : 0.73642\n",
      "precision : 0.5247466588338963\n",
      "recall : 0.26435335898194734\n"
     ]
    },
    {
     "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": [
    "from sklearn.metrics import roc_curve,auc\n",
    "#计算auc\n",
    "fpr,tpr,thresholds =roc_curve(train['Accept'],train['fitted_labels'])\n",
    "auc=auc(fpr,tpr)\n",
    "print('神经网络：')\n",
    "print('auc : '+str(auc))\n",
    "\n",
    "#以fpr假正率为横轴，tpr真征率为纵轴绘出auc曲线\n",
    "plt.plot(fpr,tpr)\n",
    "plt.plot([0,1],[0,1],'r--')\n",
    "\n",
    "from sklearn.metrics import accuracy_score,precision_score,recall_score\n",
    "#计算accuracy\n",
    "accuracy = accuracy_score(train['Accept'],train['fitted_labels'])\n",
    "print('accuracy : '+str(accuracy))\n",
    "#计算precision\n",
    "precision =precision_score(train['Accept'],train['fitted_labels'])\n",
    "print('precision : '+str(precision))\n",
    "#计算recall\n",
    "recall=recall_score(train['Accept'],train['fitted_labels'])\n",
    "print('recall : '+str(recall))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "由上，我们得出神经网络算法的auc为 0.5878284720575782"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 支持向量机SVM算法"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "相对于神经网络，支持向量机算法结构简单，性能也有明显的提升。我们将使用sklearn里的svm来训练训练集；同样，为了真实的描述模型的准确性，我们将执行K交叉验证：通过 sklearn.model_selection库的cross_val_predict() 函数来实现。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "#使用支持向量机来训练模型\n",
    "from sklearn import svm\n",
    "from sklearn.model_selection import cross_val_predict\n",
    "\n",
    "train=transform_train\n",
    "\n",
    "#设置惩罚因子，我们加重惩罚错误预测为1的\n",
    "penalty = {\n",
    "    0: 1,\n",
    "    1: 3.6\n",
    "}\n",
    "\n",
    "features =train.columns.drop('Accept')\n",
    "\n",
    "#实例化模型对象：\n",
    "model = svm.SVC(class_weight=penalty) \n",
    "model.fit(train[features],train['Accept'])\n",
    "\n",
    "#使用5重交叉验证进行预测\n",
    "fitted_labels=cross_val_predict(model,train[features],train['Accept'],cv=5)\n",
    "train['fitted_labels']=fitted_labels"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "同样，训练模型后，我们将评估所预测的训练集结果。我们使用四个指标：auc(曲线下面积），accuracy(准确率），precision(精准率）和recall(召回率），四个指标的对比，我们将在后文给出。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "支持向量机SVM：\n",
      "auc : 0.7123324850382994\n",
      "accuracy : 0.62896\n",
      "precision : 0.4137612165216796\n",
      "recall : 0.8938295353654927\n"
     ]
    },
    {
     "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": [
    "from sklearn.metrics import roc_curve,auc\n",
    "#计算auc\n",
    "fpr,tpr,thresholds =roc_curve(train['Accept'],train['fitted_labels'])\n",
    "auc=auc(fpr,tpr)\n",
    "print('支持向量机SVM：')\n",
    "print('auc : '+str(auc))\n",
    "\n",
    "#以fpr假正率为横轴，tpr真征率为纵轴绘出auc曲线\n",
    "plt.plot(fpr,tpr)\n",
    "plt.plot([0,1],[0,1],'r--')\n",
    "\n",
    "from sklearn.metrics import accuracy_score,precision_score,recall_score\n",
    "#计算accuracy\n",
    "accuracy = accuracy_score(train['Accept'],train['fitted_labels'])\n",
    "print('accuracy : '+str(accuracy))\n",
    "#计算precision\n",
    "precision =precision_score(train['Accept'],train['fitted_labels'])\n",
    "print('precision : '+str(precision))\n",
    "#计算recall\n",
    "recall=recall_score(train['Accept'],train['fitted_labels'])\n",
    "print('recall : '+str(recall))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "由上，可以得出支持向量机算法的auc值是0.7123324850382994。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 选择模型及预测测试集"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "综上，逻辑回归算法的auc值最高，为0.7175635602356261，因此我们本例中将选择该算法所生成的模型对测试集进行预测。我们选择auc作为主要指标的原因是auc指标可以无视样本目标项的不均衡分布。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "首先，我们对测试集也做如上的特征转换处理："
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Apartment</th>\n",
       "      <th>Beds</th>\n",
       "      <th>Review</th>\n",
       "      <th>Pic Quality</th>\n",
       "      <th>normalized_price</th>\n",
       "      <th>date_label</th>\n",
       "      <th>week_label</th>\n",
       "      <th>Region_1</th>\n",
       "      <th>Region_2</th>\n",
       "      <th>Region_3</th>\n",
       "      <th>Region_4</th>\n",
       "      <th>Region_5</th>\n",
       "      <th>Region_6</th>\n",
       "      <th>Region_7</th>\n",
       "      <th>Region_8</th>\n",
       "      <th>Region_9</th>\n",
       "      <th>Region_10</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>4.923078</td>\n",
       "      <td>0.683932</td>\n",
       "      <td>-0.341719</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>4.842211</td>\n",
       "      <td>0.958800</td>\n",
       "      <td>-0.230183</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>4.702331</td>\n",
       "      <td>0.500134</td>\n",
       "      <td>0.742937</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>4.388125</td>\n",
       "      <td>0.986082</td>\n",
       "      <td>-1.210650</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>3.868994</td>\n",
       "      <td>0.652980</td>\n",
       "      <td>-0.333568</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Apartment  Beds    Review  Pic Quality  normalized_price  date_label  \\\n",
       "0          1     1  4.923078     0.683932         -0.341719           1   \n",
       "1          1     1  4.842211     0.958800         -0.230183           1   \n",
       "2          1     3  4.702331     0.500134          0.742937           4   \n",
       "3          1     1  4.388125     0.986082         -1.210650           4   \n",
       "4          0     1  3.868994     0.652980         -0.333568           4   \n",
       "\n",
       "   week_label  Region_1  Region_2  Region_3  Region_4  Region_5  Region_6  \\\n",
       "0           1         1         0         0         0         0         0   \n",
       "1           1         0         0         0         0         0         0   \n",
       "2           1         0         1         0         0         0         0   \n",
       "3           1         0         0         0         0         0         0   \n",
       "4           1         0         0         1         0         0         0   \n",
       "\n",
       "   Region_7  Region_8  Region_9  Region_10  \n",
       "0         0         0         0          0  \n",
       "1         0         0         1          0  \n",
       "2         0         0         0          0  \n",
       "3         0         1         0          0  \n",
       "4         0         0         0          0  "
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#对测试集也做如上的特征转换处理\n",
    "test_original=pd.read_csv('case2_testing.csv')\n",
    "test=test_original.copy()\n",
    "\n",
    "#标准化'Price'变量\n",
    "test['normalized_price'] = (test['Price'] - test['Price'].mean()) / (test['Price'].std())\n",
    "del test['Price']\n",
    "#删去‘ID’变量\n",
    "del test['ID']\n",
    "\n",
    "\n",
    "#把日期（365天）归类转化为季度（1,2,3,4季度）\n",
    "def assign_label(h):\n",
    "    if h >=0 and h < 92:\n",
    "        return 1\n",
    "    elif h >= 92 and h <183:\n",
    "        return 2\n",
    "    elif h >=183 and h <275 :\n",
    "        return 3\n",
    "    elif h >= 275 and h < 366:\n",
    "        return 4   \n",
    "test[\"date_label\"] = test[\"Date\"].apply(assign_label)\n",
    "del test['Date']\n",
    "\n",
    "\n",
    "#将周日到周四归为类别1，周五和周六归为类别2\n",
    "def assign_week(h):\n",
    "    if h >=0 and h < 5 or h>=7:\n",
    "        return 1\n",
    "    elif h >= 5 and h <7:\n",
    "        return 2\n",
    "\n",
    "test[\"week_label\"] = test[\"Weekday\"].apply(assign_week)\n",
    "del test['Weekday']\n",
    "\n",
    "\n",
    "#将原测试集里代表类别的特征‘Region’虚拟化处理\n",
    "dum=pd.get_dummies(test['Region'],prefix=\"Region\")\n",
    "test=pd.concat([test,dum],axis=1)\n",
    "del test['Region']\n",
    "\n",
    "#显示特征转换后的前五行\n",
    "test.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Apartment</th>\n",
       "      <th>Beds</th>\n",
       "      <th>Review</th>\n",
       "      <th>Pic Quality</th>\n",
       "      <th>normalized_price</th>\n",
       "      <th>date_label</th>\n",
       "      <th>week_label</th>\n",
       "      <th>Region_1</th>\n",
       "      <th>Region_2</th>\n",
       "      <th>Region_3</th>\n",
       "      <th>Region_4</th>\n",
       "      <th>Region_5</th>\n",
       "      <th>Region_6</th>\n",
       "      <th>Region_7</th>\n",
       "      <th>Region_8</th>\n",
       "      <th>Region_9</th>\n",
       "      <th>Region_10</th>\n",
       "      <th>possibility</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>4.923078</td>\n",
       "      <td>0.683932</td>\n",
       "      <td>-0.341719</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>4.842211</td>\n",
       "      <td>0.958800</td>\n",
       "      <td>-0.230183</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>4.702331</td>\n",
       "      <td>0.500134</td>\n",
       "      <td>0.742937</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>4.388125</td>\n",
       "      <td>0.986082</td>\n",
       "      <td>-1.210650</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>3.868994</td>\n",
       "      <td>0.652980</td>\n",
       "      <td>-0.333568</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Apartment  Beds    Review  Pic Quality  normalized_price  date_label  \\\n",
       "0          1     1  4.923078     0.683932         -0.341719           1   \n",
       "1          1     1  4.842211     0.958800         -0.230183           1   \n",
       "2          1     3  4.702331     0.500134          0.742937           4   \n",
       "3          1     1  4.388125     0.986082         -1.210650           4   \n",
       "4          0     1  3.868994     0.652980         -0.333568           4   \n",
       "\n",
       "   week_label  Region_1  Region_2  Region_3  Region_4  Region_5  Region_6  \\\n",
       "0           1         1         0         0         0         0         0   \n",
       "1           1         0         0         0         0         0         0   \n",
       "2           1         0         1         0         0         0         0   \n",
       "3           1         0         0         0         0         0         0   \n",
       "4           1         0         0         1         0         0         0   \n",
       "\n",
       "   Region_7  Region_8  Region_9  Region_10  possibility  \n",
       "0         0         0         0          0            1  \n",
       "1         0         0         1          0            1  \n",
       "2         0         0         0          0            1  \n",
       "3         0         1         0          0            1  \n",
       "4         0         0         0          0            0  "
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#用前面已训练的逻辑回归模型lr对测试集进行预测\n",
    "prediction=lr.predict(test)\n",
    "test['possibility']=prediction\n",
    "\n",
    "test.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>ID</th>\n",
       "      <th>possibility</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   ID  possibility\n",
       "0   1            1\n",
       "1   2            1\n",
       "2   3            1\n",
       "3   4            1\n",
       "4   5            0"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#导出所预测的测试集结果\n",
    "test['ID']=test_original['ID']\n",
    "test_result=test[['ID','possibility']]\n",
    "\n",
    "test_result.to_csv('./test_result1.csv')\n",
    "test_result.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 指标对比：accuracy(准确率），precision(精准率）和recall(召回率）"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "accuracy(准确率）：在这里我们不用准确率作为主要指标，因为在二分类的正反比例不均衡的情况下，accuracy(准确率）的参考较低。例如本例中，训练集目标项的0所占的比例约为70%，所以即使模型全部预测结果为0，准确率仍等达到70%。训练集目标项的分布如下："
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x1eb2001f048>"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "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": [
    "transform_train['Accept'].hist()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "recall(召回率）：召回率的定义为实际为正的样本中被预测为正样本的概率。召回率越高，代表实际能租出去的房子被预测出来的概率就越高。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "precision(精准率）:精准率的含义为在所有被预测为正的样本中，实际为正的样本的概率。精准率越低，说明有较多的0样本被预测成了1，即是较多不合理价格的房子被预测为能租出去。这样就导致一个比较严重的后果：房子可能有较长时间的空置，这样无疑导致较大的损失。因为我们都知道，即使稍微便宜一点租出去，也比租不出去好。因此在本例中指标重要性方面：precision(精准率）>recall(召回率）>accuracy(准确率。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "综上：本例中指标重要性方面：precision(精准率）>recall(召回率）>accuracy(准确率）"
   ]
  }
 ],
 "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": "3.7.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}