predict_salary.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h1>预测工资</h1>\n",
    "\n",
    "既然我们有了拉勾的数据，我们可以用它来耍耍，预测一下工资。\n",
    "\n",
    "\n",
    "<h3>1、对工资进行转换</h3>\n",
    "\n",
    "在这之前先导入模块并读入文件，不仅有训练数据文件，还有一组自拟的测试数据文件。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((450, 3), (24, 2))"
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "train_file = \"analyst.csv\"\n",
    "test_file = \"test.csv\"\n",
    "# 读取文件获得数据\n",
    "train_data = pd.read_csv(train_file, encoding=\"gbk\")\n",
    "train_data = train_data.drop('ID', axis=1)\n",
    "\n",
    "test_data = pd.read_csv(test_file, encoding=\"gbk\")\n",
    "train_data.shape, test_data.shape\n",
    "# data['薪资'].value_counts()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "为了更好地进行分析，我们要对薪资做一个预处理。由于其分布比较散乱，很多值的个数只有1。为了不造成过大的误差，根据其分布情况，可以将它分成【5k 以下、5k-10k、10k-20k、20k-30k、30k-40k、40k 以上】，为了更加方便我们分析，取每个薪资范围的中位数，并划分到我们指定的范围内。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "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>学历要求</th>\n",
       "      <th>工作经验</th>\n",
       "      <th>薪资</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>大专</td>\n",
       "      <td>3-5年</td>\n",
       "      <td>10k-20k</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>本科</td>\n",
       "      <td>3-5年</td>\n",
       "      <td>10k-20k</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>本科</td>\n",
       "      <td>不限</td>\n",
       "      <td>5k-10k</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>本科</td>\n",
       "      <td>5-10年</td>\n",
       "      <td>30k-40k</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>本科</td>\n",
       "      <td>3-5年</td>\n",
       "      <td>30k-40k</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  学历要求   工作经验       薪资\n",
       "0   大专   3-5年  10k-20k\n",
       "1   本科   3-5年  10k-20k\n",
       "2   本科     不限   5k-10k\n",
       "3   本科  5-10年  30k-40k\n",
       "4   本科   3-5年  30k-40k"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "salarys = train_data['薪资'].unique()     # 获取到薪资的不同值\n",
    "for salary in salarys:\n",
    "    # 根据'-'进行分割并去掉'k',分别将两端的值转换成整数\n",
    "    min_sa = int(salary.split('-')[0][:-1])\n",
    "    max_sa = int(salary.split('-')[1][:-1])\n",
    "    # 求中位数\n",
    "    median_sa = (min_sa + max_sa) / 2\n",
    "    # 判断其值并划分到指定范围\n",
    "    if median_sa < 5:\n",
    "        train_data.replace(salary, '5k以下', inplace=True)\n",
    "    elif median_sa >= 5 and median_sa < 10:\n",
    "        train_data.replace(salary, '5k-10k', inplace=True)\n",
    "    elif median_sa >= 10 and median_sa < 20:\n",
    "        train_data.replace(salary, '10k-20k', inplace=True)\n",
    "    elif median_sa >= 20 and median_sa < 30:\n",
    "        train_data.replace(salary, '20k-30k', inplace=True)\n",
    "    elif median_sa >= 30 and median_sa < 40:\n",
    "        train_data.replace(salary, '30k-40k', inplace=True)\n",
    "    else:\n",
    "        train_data.replace(salary, '40k以上', inplace=True)\n",
    "train_data.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "处理完成后，我们可以将“薪资”单独提取出来当作训练集的 label。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_train = train_data.pop('薪资').values"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h3>2、对变量进行转换</h3>\n",
    "\n",
    "<h4>把category的变量转变成numerical表达式</h4>\n",
    "\n",
    "由于变量都不是numerical变量，在训练的时候计算机没办法识别，因此要对它们进行转换。\n",
    "\n",
    "当我们用numerical来表达categorical的时候，要注意，数字本身有大小的含义，所以乱用数字会给之后的模型学习带来麻烦。于是我们可以用One-Hot的方法来表达category。\n",
    "\n",
    "pandas自带的get_dummies方法，可以一键做到One-Hot。\n",
    "\n",
    "这里按我的理解解释一下One-Hot：比如说data['学历要求']有'大专', '本科', '硕士', '不限'。但data['学历要求']=='本科'，则他可以用字典表示成这样{'大专': 0, '本科':1, '硕士':0, '不限':0}，用向量表示为[0, 1, 0, 0]\n",
    "\n",
    "在此之前，将测试集和训练集组合起来一起处理，稍微方便一点。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(474, 2)"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data = pd.concat((train_data, test_data), axis=0)\n",
    "data.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "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>学历要求_不限</th>\n",
       "      <th>学历要求_大专</th>\n",
       "      <th>学历要求_本科</th>\n",
       "      <th>学历要求_硕士</th>\n",
       "      <th>工作经验_1-3年</th>\n",
       "      <th>工作经验_1年以下</th>\n",
       "      <th>工作经验_3-5年</th>\n",
       "      <th>工作经验_5-10年</th>\n",
       "      <th>工作经验_不限</th>\n",
       "      <th>工作经验_应届毕业生</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\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>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\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>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\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>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\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>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\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>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   学历要求_不限  学历要求_大专  学历要求_本科  学历要求_硕士  工作经验_1-3年  工作经验_1年以下  工作经验_3-5年  \\\n",
       "0        0        1        0        0          0          0          1   \n",
       "1        0        0        1        0          0          0          1   \n",
       "2        0        0        1        0          0          0          0   \n",
       "3        0        0        1        0          0          0          0   \n",
       "4        0        0        1        0          0          0          1   \n",
       "\n",
       "   工作经验_5-10年  工作经验_不限  工作经验_应届毕业生  \n",
       "0           0        0           0  \n",
       "1           0        0           0  \n",
       "2           0        1           0  \n",
       "3           1        0           0  \n",
       "4           0        0           0  "
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dummied_data = pd.get_dummies(data)\n",
    "dummied_data.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "上次可视化分析的时候就已经知道数据集中不存在缺失值了，为了走一下流程并确保正确性，再次看一下是否有缺失值。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "工作经验_应届毕业生    0\n",
       "工作经验_不限       0\n",
       "工作经验_5-10年    0\n",
       "工作经验_3-5年     0\n",
       "工作经验_1年以下     0\n",
       "工作经验_1-3年     0\n",
       "学历要求_硕士       0\n",
       "学历要求_本科       0\n",
       "学历要求_大专       0\n",
       "学历要求_不限       0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dummied_data.isnull().sum().sort_values(ascending=False).head(10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "OK，很好，没有缺失值。这些值比较简单，不需要做那么多工作。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h2>3、建立模型</h2>\n",
    "\n",
    "<h4>DecisionTree</h4>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(450, 24)"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train = dummied_data[:train_data.shape[0]].values\n",
    "X_test = dummied_data[-test_data.shape[0]:].values\n",
    "len(X_train), len(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.tree import DecisionTreeClassifier\n",
    "from sklearn.model_selection import cross_val_score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0xd6ef9b0>]"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "features_scores = []\n",
    "max_features = [.1, .2, .3, .4, .5, .6, .7, .8, .9]\n",
    "for max_feature in max_features:\n",
    "    clf = DecisionTreeClassifier(max_features=max_feature)\n",
    "    features_score = cross_val_score(clf, X_train, y_train, cv=5)\n",
    "    features_scores.append(np.mean(features_score))\n",
    "plt.plot(max_features, features_scores)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "可见当 max_features = 0.2 时达到最大，大概有0.5418\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "dtc = DecisionTreeClassifier(max_features=0.2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h3>ensemble（集成算法）</h3>\n",
    "\n",
    "集成学习简单理解就是指采用多个分类器对数据集进行预测，从而提高整体分类器的泛化能力。\n",
    "    \n",
    "<h4>AdaBoostClassifier</h4>\n",
    "\n",
    "adaptive boosting: 通过改变训练样本的权值，学习多个分类器，并将这些分类器进行线性组合，提高泛化性能。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.ensemble import AdaBoostClassifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0xd806450>]"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYQAAAD8CAYAAAB3u9PLAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvhp/UCwAAIABJREFUeJzt3XlcVPe98PHPl2ETBBGBAYKKiiKiYCJZmlUjSTQRk+d2S7qk6W2epFlu2vs8t7fJfZK0Sdrb295utzdpb02aNkmXtDdtEzEmGo1LNmPcQFBRwAUFZBNBkGWY3/MHx16CKCPMzJlhvu/Xa17OnDnLlyPMd873/BYxxqCUUkqF2R2AUkqpwKAJQSmlFKAJQSmllEUTglJKKUATglJKKYsmBKWUUoAmBKWUUhZNCEoppQAPE4KILBGRChGpFJGHh3j/LhFpFJFd1uPuQe/Hi8gxEXl6wLKN1j7PbJMy+h9HKaXUSIUPt4KIOIBngBuAo8BHIrLSGLNn0Kp/NMY8eI7dPAVsGmL5540x2zwNNikpyWRmZnq6ulJKKWD79u1Nxpjk4dYbNiEAlwGVxphqABF5GbgVGJwQhiQiCwAn8CZQ4Mk255KZmcm2bR7nD6WUUoCIHPZkPU9KRhcBNQNeH7WWDfZJESkVkVdEZLIVRBjwI+Ab59j3r61y0WMiIp4ErJRSyjc8SQhDfVAPHhGvGMg0xuQB64AXrOX3A6uNMTWc7fPGmHnANdbji0MeXOQeEdkmItsaGxs9CFcppdRIeJIQjgKTB7zOAGoHrmCMaTbGdFsvnwUWWM8/ATwoIoeAHwJ3isi/Wdscs/5tB35Pf2nqLMaYFcaYAmNMQXLysCUwpZRSI+TJPYSPgJkiMg04BtwOfG7gCiKSZoyps14uB/YCGGM+P2Cdu4ACY8zDIhIOJBhjmkQkAlhG/5WFUkopmwybEIwxLhF5EFgDOIDnjTHlIvIksM0YsxJ4SESWAy6gBbhrmN1GAWusZOCgPxk8O/IfQyml1GhJME2QU1BQYLSVkVJKXRgR2W6MGbaVp/ZUVkopBWhCUCGks8fFS1sO0+Ny2x2KUgHJk5vKSo0JP1q7n1+9e5CYCAefXJBhdzhKBRy9QlAhYU9tG795/xAAxaW1519ZqRClCUGNeW634dFXdzNhXAS3XzqZdw800dLRY3dYSgUcTQhqzPvv7TXsONLKI0tn84UrpuJyG94sq7c7LKUCjiYENaa1dPTwvTf2cVlmIp9akEFuejzTk2IpLtGykVKDaUJQY9r339hHe5eLp26bi4ggIizLT2fLwWYa2rrsDk+pgKIJQY1Z2w+38MdtNXzl6mlkp8b9bfny/DSMgVWldefZWqnQowlBjUmuPjf/769lpE2I5muLZ37svayUOHLS4rW1kVKDaEJQY9ILHxxmX3073yqaQ2zU2d1tivLT2HmklZqWThuiUyowaUJQY079yS5+vLaChdnJ3JSbOuQ6RXnpgJaNlBpIE4Iac556fQ8ut+HJ5f03kocyOTGG+ZMTtLWRUgNoQlBjyub9jbxeWscDi7KYMinmvOsW5aezp66NyoZTfopOqcCmCUGNGV29fTz+WhnTkmK597rpw66/LC8NEfQqQSmLJgQ1ZvxyUzWHmjt58tZcosIdw67vjI/m8mmJFJfWEkzzgijlK5oQ1JhwuLmDZzZWsiwvjWtmej73dlF+OtWNHeypa/NhdEoFB00IKugZY3j8tXIiHWE8tmzOBW27dG4ajjChuERbGymlCUEFvTfL6tm0v5F/vGEWzvjoC9o2MTaSq7OSKC7RspFSmhBUUDvV7eKJ4j3kpMXzpU9MHdE+ivLTOdZ6mp01rV6OTqngoglBBbX/WLef+rYuvnPbXMIdI/t1vjHXSWR4mLY2UiFPE4IKWvvq23j+vUPcfulkFkydOOL9xEdHsCg7mVWldfS5tWykQpcmBBWU3G7Do38tIz46nG8umT3q/RXlp9PY3s2HB5u9EJ1SwUkTggpKr+w4yrbDJ3hkaQ4TYyNHvb/rZ6cQE+nQ1kYqpGlCUEHnREcP31u9lwVTJ/KpBRle2WdMZDiFOU7eKKujt8/tlX0qFWw0Iaig84M1FbR1ufjObXMJCxt68LqRKMpPp7Wzl3crm7y2T6WCiSYEFVR2HDnByx8d4ctXZpKTFu/VfV87K4n46HBtbaRCliYEFTRcfW4e/WsZzrhovn7DLK/vPyrcwZK5qawtP05Xb5/X969UoNOEoILGS1sOs6eujceL5jB+iFnQvKEoP51T3S42VjT4ZP9KBTJNCCooNLR18aO1+7l2VjJL5w49C5o3fGL6JCbFRmprIxWSNCGooPDU63vp6XPz5PLcc86C5g3hjjBunpfG+n3HOdXt8tlxlApEHiUEEVkiIhUiUikiDw/x/l0i0igiu6zH3YPejxeRYyLy9IBlC0Rkt7XPn4kv/8pVUHv3QBPFJbXcv3AGmUmxPj9eUX46Xb1u1u897vNjKRVIhk0IIuIAngGWAnOAO0RkqDGG/2iMmW89nhv03lPApkHLfgHcA8y0HksuNHg19nW7+mdBmzophq9eN8MvxyyYOpG0CdHa2kiFHE+uEC4DKo0x1caYHuBl4FZPDyAiCwAnsHbAsjQg3hjzgekfc/hF4LYLilyFhBWbqqlu6uCJ5blERww/C5o3hIUJy/LS2LS/kZOdvX45plKBwJOEcBFQM+D1UWvZYJ8UkVIReUVEJgOISBjwI+AbQ+zzqAf7VCHsSHMnT2+o5OZ5qSzMTvHrsYvy0+ntM7xZrjeXVejwJCEMVdsfPCRkMZBpjMkD1gEvWMvvB1YbY2oGre/JPvtXFLlHRLaJyLbGxkYPwlVjgTGGb60sIzxMeHxZrt+PP++iCUydFKOtjVRI8SQhHAUmD3idAXysuGqMaTbGdFsvnwUWWM8/ATwoIoeAHwJ3isi/WfscOAjNWfscsO8VxpgCY0xBcrLnc+Wq4Lam/DgbKvpnQUudcGGzoHmDiFCUl877VU00tncPv4FSY4AnCeEjYKaITBORSOB2YOXAFax7AmcsB/YCGGM+b4yZYozJBP4JeNEY87Axpg5oF5ErrNZFdwKvjf7HUWNBR7eLJ4vLmZ0ax5euzLQtjqL8dNwG3ijTq4ThuPrcvHOgEbfOJxHUhk0IxhgX8CCwhv4P+j8ZY8pF5EkRWW6t9pCIlItICfAQcJcHx74PeA6oBKqAN0YQvxqDfvb2AWpP9s+CFjHCWdC8ITs1jlnO8drayAMr3qnmi7/ayktbDtsdihoFj/r/G2NWA6sHLXt8wPNHgEeG2cdvgN8MeL0NmOt5qCoU7D/ezq/eOchnCjIoyEy0OxyW56fzw7X7qW09TXrCOLvDCUine/r41TsHAfjh2gqWzkslJc7/ZT41etpTWQUMYwyPvlrG+OhwHl6aY3c4ACzLSwdgValeJZzLH7Yeobmjhx98Mo/uXjffW73P7pDUCGlCUAHjLzuOsfVgCw8vmU2iF2ZB84bMpFjyMiZoa6Nz6Hb1sWJzNZdNS+Qzl07m3uum89edx9hSrVORBiNNCCognOzs5V9X7+XiKQl8pmDy8Bv4UVFeOruPneRgU4fdoQScP28/Rn1bFw8uygLg/oVZZEwcx2OvlunMc0FIE4IKCD9Ys48TnT1enwXNG27J629Et0pvLn+Mq8/NLzZVkp8xgWtmJgEwLtLBE8tzOdBwiuffPWhzhOpCaUJQtttV08rvtx7hriunkZs+we5wzpKeMI5LMydSrPcRPmZlSS01Lad5YFHWx0agXZzjpDDHyU/XHaC29bSNEaoLpQlB2arPbXj01d0kj4/iH2+YaXc457Q8P539x09RUd9udygBwe02PLOhktmpcRTmOM96/1tFczAYnlq1x4bo1EhpQlC2+u2Ww5Qda+OxZXOIi46wO5xzWjovjTBB+yRY3iyvp6qxg/sXZQ1Z4pucGMM/XD+TN8rqdfa5IKIJQdmmob2LH66p4JqZSSzLSxt+AxsljY/iqqwkVpbU0j9Ab+gypv/qYFpSLLfMO/f/293XTGN6cizfWlmuc1QHCU0Iyjb/+vpeul1unvDxLGjeUpSXzpGWTkqPnrQ7FFttrGikvLaN+xbOwHGeBgBR4Q6eunUuh5s7+eWmaj9GqEZKE4KyxfuVTby6q5avXjed6cnj7Q7HIzflphLhkJAuGxlj+M+3D3BRwjj+18XDj1h/VVb/1d8zGys53KzNdgOdJgTldz0uN4++VsaUxBjut9qvB4MJMRFcNyuZVaV1ITuI2wfVzew40spXr5vu8ThTjy2bQ6QjjG+vLA/5clug04Sg/O7Zd6qpbuzgiVv9NwuatxTlp1Pf1sW2wyfsDsUWT79dSXJcFJ++gM6Dzvhovl44kw0Vjawp13mqA5kmBOVXNS2d/OfbB1iSm8oiP8+C5g2FOU6iI8JCsmy0/fAJ3q9q5p5rpl9wIr/rykxmp8bxZHE5nT0uH0WoRksTgvKrJ4rLCRPh8aI5docyIrFR4SzOcbJ6dx2uEBua4ZkNlSTERPC5y6dc8LbhjjC+c9tcak928bP1lT6ITnlDyCQErV3a7609x1m3t4GvF84M6qGki/LSae7o4f2q0BnArbz2JG/va+ArV00jNsqjUfPPUpCZyKcXZPDcO9UcOK4d/C5ETUunX44TEgnhm6+Ucu9L2+0OI6R19rj49spyZjnH8+WrptkdzqgszE5mfFR4SJWNfr6hiriocO4c5Qx2Dy+dTWxUOI+9VqZf0jz0h61HWPyjTZTUtPr8WCGREGKjwtm4v1Frlzb6z7crOdZ6mu/cNs/WWdC8ITrCwY25Tt4sr6fbNfY7XFU2tLO6rI47r5zKhHGj600+aXwU37gpmy3VLawMoYQ6UtsPn+Dx18q4YsYk5l7k+3G+gvsv00OFOSn0uNy8e6DJ7lBC0oHj7Ty7uZpPLcjgsmn2z4LmDUX56bR3udi8f+z/Tv18YxXR4Q7+3ktXdndcNoX8jAl85/W9tHX1emWfY9Hxti7u++120iaM42e3zz9vJ0BvCYmEcOm0ROKiw1m/V8dU8TdjDI+9VkZsVDiPLJ1tdzhec3VWEhNjIsZ82ehIcyev7arlc5dPYdL4KK/s0xEmPHXbXJpOdfPjtfu9ss+xptvVx32/3c6pbhcr7lxAQox/JowKiYQQ4QjjulnJrN/XELIdiuzy6q5jbKlu4Z+XZHvtAyUQRDjCWDovjbf2HB/Tpcj/2lyFQ4R7rp3u1f3mZSTwhcun8uIHhyivDe2hQIby7ZXl7DjSyg8/nc/s1Hi/HTckEgLADXOcNJ3qpuSo72/MqH4nT/fy3df3kj85gTsuvfCmioGuKC+d0719Y/bKs/5kF69sO8qnCjJwxkd7ff//dGM2E2MiefTVMv2iNsDvPjzMH7bWcP/CGdx8nsEDfSFkEsLCWSk4wmTM/vEGoh+traClo4fvBuAsaN5w2bREUuKixmzZaMXmavqM4b7rZvhk/xNiIviXm3PYeaSV/95e45NjBJtth1r49spyFmYn839vzPb78UMmIUyIiaBg6kTW7dWu8/5QerSVl7Yc5s5PZPqldYQdHGHCLXlpbKxoHHM3R5tOdfP7rYe5dX46kxNjfHacv7vkIi7LTOTf3tjHiY4enx0nGNSf7OK+3+0gPWEc//HZi/1yE3mwkEkI0F822lffztET/unkEar6Z0ErI2l8FP/nxll2h+NTRfnp9PS5WTvGxuh5/t2DdLvc3L/Qt4MPighP3pZLW5eLH6zZ59NjBbJuVx9f/e12OrpdrPhiARNi7JksKqQSwmJrqj8tG/nW77ceofToSR69JYf4AJ4FzRsunpxAxsRxY6psdLKzlxc/OMzNc9PISvH90OSzU+P5+6syefmjGnYcCb1BA40xPP5qObtqWvnxZ/LJTo2zLZaQSgjTkmKZnhyrZSMfamzv5gdv7uPKGZNYnp9udzg+JyIU5afzbmUTLWOk5PHCB4c41e3iAT8OTf61wlmkxEXx2Ktl9IXYDebffniEP26r4cFFWSyZa+/MgSGVEKB/tMot1c20j7Gab6D43uq9dPX28eStc4NiFjRvKMpLp89tWL27zu5QRq2j28Xz7x1k8ewU5qT7r7nj+KhwHl+WS3ltG7/dcthvx7XbR4daeGJlOYuyk/nHG+wvr4ZkQujtM7yjvZa9bkt1M3/ZeYx7r53hl1JDoMhJi2NGcuyYKBv97sPDtHb28sD1/p+46OZ5qVwzM4kfrqmgob3L78f3t7qTp7nvtzuYnBjDT2+35ybyYCGXEC6ZkkBCTISWjbysx+XmsVfLyJg4zq+lhkBwpmy09VAL9SeD94Osq7ePZ985yFVZk7hkykS/H19EeGJ5Lt0uN99bPbZvMHf19vHVl7ZzusfFii8uGPUYUd4Scgkh3BHGouwUNuxrCLlapS/96t2DHGg4xRPLcxkXGVyzoHlDUX46xsDrQVw2+tO2Ghrbu21N6NOTx3PvddP5685jfDBGhxc3xvDYq2WUHD3Jjz87n5lO+24iDxZyCQH6y0YnOntDskWDLxw90cnP1h/gxjnOv7XkCjUzkseTmx4ftGWj3j43v9xUzYKpE/nE9Em2xnL/wiwyJo7j8dfK6HGNvUmIXtpymP/efpSHrs/iptxUu8P5mJBMCNfOSiLCIVo28pIni/cABO0saN5SlJ/OrppWv01m4k1/3XmMY62neXBRlu2NAcZFOnhieS4HGk7x/HsHbY3F2z6sbubJ4j0snp3C1wvtv4k8mEcJQUSWiEiFiFSKyMNDvH+XiDSKyC7rcbe1fKqIbLeWlYvIVwdss9Ha55lt/DbBblx0BJdPm6T9Ebxg/d7jrN1znIcWzyRjou96tAaDW6xxZ4pLg+sqoc9t+MXGKnLT41mYnWx3OEB/n6HCHCf/se4Ata2n7Q7HK2pbT/PA73cwJTGGn9w+PyCHcxk2IYiIA3gGWArMAe4QkaG+Cv7RGDPfejxnLasDrjTGzAcuBx4WkYGN0z8/YBu/fjovzkmhsuEUh5o6/HnYMefHb+1nRnIsX7k6uGdB84bJiTFcMiWBlbuCKyG8vruOg00dAXF1MNC3iuZgMDy1ao/doYxaV29/T+SuXjcr7lwQsB02PblCuAyoNMZUG2N6gJeBWz3ZuTGmxxjTbb2M8vB4flFo1bq1bDRyVY2nKK9t43OXTyUyPGD+a21VlJ/Ovvr2oJkz2O02PPN2JVkp4wOunj05MYZ/uH4mb5TVs7EieK/mjTH8v7+WUXr0JD/57HyyUgLnJvJgnvwVXwQMHIrwqLVssE+KSKmIvCIik88sFJHJIlJq7eP7xpiBX59+bZWLHpNzfDURkXtEZJuIbGtsbPQgXM9MTowh2xmnZaNRWFVSh8j/lEpU/7kIEyguDY7WRuv2HqfieDsPLJoRkCWMu6+ZxvTkWL61spyu3uCcrvSF9w/x5x1H+drimdwwJ7AbXXiSEIb6LRncXrMYyDTG5AHrgBf+tqIxNdbyLOBLInLmjHzeGDMPuMZ6fHGogxtjVhhjCowxBcnJ3q1vLs5JYeuhFk52aq/lC2WMYWXJMS7LTCR1gvfHyg9WKfHRXDF9EqtKagN+EnljDE9vqGRKYgxFeYE5zEhUuIOnbp3L4eZO/mtTld3hXLAt1c089fpeCnOcfG3xTLvDGZYnCeEoMHnA6wzgY0VSY0zzgNLQs8CCwTuxrgzK6f/wxxhzzPq3Hfg9/aUpvyqc46TPbdi4X68SLtS++naqGjsoCoHxii5UUX461U0dlNe22R3Keb1zoInSoye5b+EMwh2BW/K7KiuJZXlp/HxjFYebg+ee37HW0zzwux1kTorhJ5/ND8grsME8+S34CJgpItNEJBK4HVg5cAURGVgzWA7stZZniMg46/lE4CqgQkTCRSTJWh4BLAPKRvvDXKj5GQkkjY/UstEIFJfU4ggTv8/oFAyW5KYSHiYB39ro6bcrSY2P5u8uGaoCHFgeWzaHSEcY31pZHvBXXtB/E/nel7bR43Kz4s4C4gL0JvJgwyYEY4wLeBBYQ/8H/Z+MMeUi8qSILLdWe8hqVloCPATcZS3PAT60lm8CfmiM2U3/DeY11r2FXcAx+q8s/CosTFiUncLGigZ6+8ZeBxhfMcZQXFrL1VlJJMb6Z/LvYDIxNpJrZiaxqqQuYKeG3Hqwha2HWrj3uulEhQd+z3JnfDRfL5zJxopG1gT43BPGGP7lL7spO9bGTz47nxnJwTOul0fXicaY1caYWcaYGcaY71rLHjfGrLSeP2KMyTXG5BtjFhlj9lnL3zLG5FnL84wxK6zlHcaYBdayXGPM14wxttwxWpzjpK3LxUeHWuw4fFAqOXqSmpbTWi46j6L8dI61nmZnTWD2hn96QyWTYiO5PYjmur7rykxmp8bxZHE5nT0uu8M5p1+/d4i/7DzGPxbOojDAbyIPFriFQz+5ZmYSkeFhWja6AMUltUQ6wrgxN7h+2f3phjlOosLDKC4JvNZGJTWtbN7fyN3XTA+qcafCHWF857a51J7s4mfrK+0OZ0jvVzXx3dV7uXGOk3+wYcTY0Qr5hBAbFc6VMyaxbu/xoKhN2s3tNqwqrWVhdnLAdq4JBHHREVw/O4VVpXUBN4jiMxsqiY8O5wtXBM/VwRkFmYl8ekEGz71THXB9PY6e6OTB3+9kWlIsP/pMcNxEHizkEwL0l40ON3dS1XjK7lAC3keHWjje1q3lIg8U5afTdKqbD6sDZ9TOffVtrN1znC9fNS1obnQO9vDS2cRGhfPYa2UB8yXudE8f9760nd4+Nyu+uCBoz60mBKAwp38YpXVaNhpWcWkt4yIcLM7x29BTQWtRdgqxkY6Aam308w1VxEY6+PJVmXaHMmKTxkfxjZuy2VLdwsoAGF3WGMMjfyllT10b/3H7fKYH0U3kwTQhAGkTxpGbHs96HcbivFx9blbvrqdwjpOYyHC7wwl44yId3DDHyRtl9QExjPPBpg5WldbyhU9MJSEmuFuH3XHZFPIzJvDUqr202Twd7q/ePciru2r5P4WzuH52cN9X04RgWZzjZPvhE2NmonRfeL+qmZaOHorytO+Bp4ry02nt7OXdSu8NuzJSv9hYSYQjjLuvnm53KKPmCBOeum0uzR3d/HjtftvieK+yie+9sY+bcp1jYqZATQiWwpwU3AY27NOy0bkUl9QSFx3OdQEyRHIwuGZmMhPGRdje2uhY62n+suMYt186meS4KFtj8Za8jAS+cPlUXvzgEGXHTvr9+DUtnTz4+x1MT4rlR58JzOGsL5QmBMvc9Ak446NYv0/LRkPpdvXxZnk9N+WmBkVHpkARGR7G0rmprC2vt3Vwtl9uqkIE7rluhm0x+MI/3ZjNxJhIHnutzK+dAM/cRHa5DSvuLGB81NgooWpCsISFCdfPdrJ5fxPdruAcVdGXNu9vor3Lpa2LRqAoP52Onj7brj4b2rt4+aMa/u7iDC5KGGdLDL4yISaCf7k5h51HWvnTtprhN/ACYwzf/HMpe+vb+NkdFzMtKdYvx/UHTQgDFOakcKrbxYfV2mt5sOKSWhJjI7lyhr3z7QajK6ZPIml8lG2tjZ575yCuPjf3LRxbVwdn/N0lF3FZZiLff3MfJ/xwD/C5dw6ysqSWf7oxm0XZY6u1nSaEAa7KSiI6IkxbGw3S2ePirT3HWTo3lYgAHhUzUDnChFvmpbJ+bwOnuv075MKJjh5+u+UwRfnpZI6hb7IDifTfYG7rcvGDNft8eqx3DzTxvTf2cvO8VO4fgwlW/7oHiI5wcHVWMuv2NgRMh5dA8Pa+Bk739mm5aBSK8tPpdrl5a0+9X4/76/cO0tnTNyZawJxPdmocf39VJn/YWsOOI74ZP6qmpZMH/7CDmSlx/Pun8gNqulFv0YQwSGFOCsdaT7OvPrC6xdupuKQWZ3wUl2Ym2h1K0LpkykTSJ0T7tbVRW1cvv3n/EDflOpnlDNxpG73la4WzcMZH8dirZbi8PHpxZ4+L//3iNtxuw4o7FxA7Rm4iD6YJYZDrZ/fXBLVs1K+tq5cNFY3cMi8dxxhoVmeXsDBhWX46m/c30trpn74uL31wmLYuFw8uCvyZurxhfFQ4jy/Lpby2jd9uOey1/Rpj+OdXSqk43s7P7riYqZPGZukNNCGcJSU+mvzJCTqMheWt8uP0uNwU5WtntNFanp+Oy214s8z3ZaPTPX08/+5BrpuVzLyMCT4/XqC4eV4q18xM4kdr99PQ3uWVfa7YXM2q0jq+cVM2C8fYTeTBNCEMoXB2CiVHW732CxXMiktryZg4jvmTE+wOJejlpsczLSnWL62N/rD1CM0dPTwYhEMwj4aI8MTyXLpdbv719b2j3t/m/Y18/8193DIvjfvGWB+OoWhCGMLiHCdGey3T0tHDuweaKMpPH5M30PxNRCjKS+ODqmafftnodvXxy81VXD4tMSTv+0xPHs+9103n1V21fFA18pFmDzd38A9/2MksZxz//um8kPgb0IQwhJy0OC5KGBfyZaM3y+pxuQ1Fedq6yFuK8tNxG3hjt+/KRn/efozjbd0hd3Uw0AOLspicOI7HXisb0cCCnT0u7n1pOwArvlgQMoM5akIYgoiwOCeFdw802TrcgN1WlhxjRnIsOWljv4WKv8x0xjE7Nc5nwza7+tz8YlMl+ZMTuDorySfHCAbREQ6+XZRLZcMpnn/v4AVta4zhG6+Usv94O/95x8VMmRTjoygDjyaEc1ic4+R0bx/vVzXZHYotjrd18eHBFi0X+UBRfjrbD5/g6IlOr+97ZUktNS2neXBRVsj/vy3OcVKY4+Q/1h2gtvW0x9v916ZqXi+t45tLZnPtrNAayFETwjlcMT2R2EhHyJaNXi+twxhYpuUirztTgnu91Lt9EtxuwzMbKpmdGsfi2WO7NYynvlU0B4PhyeI9Hq2/saKBH6zZR1F+OvdcG/zDhF8oTQjnEBXu4NpZybwdor2Wi0trmZMWT1ZK8M7+FKimTIohf3KC11sbvVleT1VjBw8syhoTQzF7w+TEGP7h+pm8WV7Phorzf7k71NTBQ3/YSbYzju9/cl5IXmFpQjiPxTlO6tu6KK9tszsUv6pp6WTnkVYdqsKHivIm/fiQAAAYE0lEQVTSKDvWRrWX5vE2xvD025VMT4rl5nnaZ2Sgu6+ZxvTkWL69svyc9wQ7ul3c89I2wsKEZ+8MnZvIg2lCOI9F2cmIwFt7QqvX8iqrlLFMZ0bzmWV56Yj8z7kerQ0VDeypa+O+hTO0R/kgUeEOnrp1LoebO/mvTVVnvd9/E7mEyoZTPH3HJUxODJ2byINpQjiPSeOjWDBlYshNmlNcUsvFUxJC+g/D11InRHNpZiIrS2pHXZI0xvCfb1dyUcI4brv4Ii9FOLZclZVEUX46P99YxeHmjo+99/ONVazeXc8jS3O4embotswCTQjDWpzjpOxYG3UnPW+lEMwqG06xp65N+x74QVF+OpUNp0Y9kOIHVc3sPNLKVxfO0OHJz+PRW3KIdITxrZXlf0vCG/Y18MO1Fdw6P527r5lmc4T209+eYRTmnBnsLjRaG60qrUUEbtFykc/dPDcVR5hQPMo+CU9vqCQlLopPL8jwUmRjkzM+mq8XzmRjRSNryo9zsKmDh17eSU5qPP/2d6HRE3k4mhCGkZUynimJMSEx+qkxhuKSWi6flogzPtrucMa8SeOjuCorieLSkZeNth8+wftVzdxz7XSiI3Su6+HcdWUms1PjeLK4nHte3EZ4mPDLLy5gXKSeO9CEMCwRoTDHyXtVzXT2+He2K3/bW9dOVWOHti7yo6K8NGpaTlNy9OSItn9mQyUTYyL43OVTvBzZ2BTuCOM7t82l9mQX1U0dPP250L6JPJgmBA8U5qTQ43Lz7oGx3Wu5uLQWR5iwdK6Wi/zlxtxUIh1hIyoblR07ydv7GvjK1dNCtpnkSBRkJvLE8lx++tn5XBXCw3sMRROCBy6dlkhcdDjrxnDZ6Ey56OqsJBJjI+0OJ2RMGBfBddnJrCqtxe2+sLLRzzdWEhcVzhc/kemb4MawL12ZqVfCQ/AoIYjIEhGpEJFKEXl4iPfvEpFGEdllPe62lk8Vke3WsnIR+eqAbRaIyG5rnz+TAL6jE+EI47pZyby9r/GC/2iDxa6aVo6eOK1/JDYoyk/neFs3Ww+1eLxNZUM7b5TV86UrM5kwLsKH0alQMmxCEBEH8AywFJgD3CEic4ZY9Y/GmPnW4zlrWR1wpTFmPnA58LCInPnE+QVwDzDTeiwZ3Y/iWzfMcdJ0qpuSo612h+ITxSV1RDrCuDHXaXcoIacwJ4VxEY4LKhv9fEMV0eEO/v5qbSqpvMeTK4TLgEpjTLUxpgd4GbjVk50bY3qMMd3Wy6gzxxORNCDeGPOB6W9e8SJw2wVH70cLZ6XgCJMxWTbqcxtWldayMDuZ+Gj9tulvMZHhFM5x8kZZPb0eTA5/pLmT10pq+fzlU7S8p7zKk4RwEVAz4PVRa9lgnxSRUhF5RUQmn1koIpNFpNTax/eNMbXW9kc92GfAmBATQcHUiWOyP8JHh1poaO/WcpGNivLSaOno4X0PZvj6xaYqHCL87xAcjVP5licJYaja/uBCejGQaYzJA9YBL/xtRWNqrOVZwJdExOnhPvsPLnKPiGwTkW2NjY0ehOs7N8xxsq++3Sfj2NupuKSWcREOFufokMl2uS47mbjo8GHLRnUnT/Pn7Uf5dEGG9hVRXudJQjgKTB7wOgP42G+tMaZ5QGnoWWDB4J1YVwblwDXWPgd2qzxrnwO2W2GMKTDGFCQn2ztZxeKc/vr6WLpK6O1z80ZZPYVznNp00UZR4Q5uyk1lTVk93a5zz9K3YnM1fcbw1RCY8F35nycJ4SNgpohME5FI4HZg5cAVrHsCZywH9lrLM0RknPV8InAVUGGMqQPaReQKq3XRncBro/5pfGxaUizTk2PH1H2E96uaaenooUiHqrBdUX467d0uNlYMfSXcdKqbP2w9wm3zL9LOVMonhk0IxhgX8CCwhv4P+j8ZY8pF5EkRWW6t9pDVrLQEeAi4y1qeA3xoLd8E/NAYs9t67z7gOaASqALe8NLP5FOFOU62VDfT3tVrdyheUVxSS1x0ONdlh9ZUgYHoqhmTSIyNPGfZ6Pl3D9LtcnP/Ir06UL7hUY3AGLMaWD1o2eMDnj8CPDLEdm8BeefY5zZg7oUEGwgKc5ys2FzNOweagn4ikm5XH2vK6rlpbipR4TqWi93CHWHcPC+VP28/RmeP62MlvJOdvbz4wWFunpfGjGSdxU75hvZUvkCXTEkgISaCdWNg0pxNFY20d7u0dVEAKcpL53Rv31lzeb/wwSFOdbt4cFGWPYGpkKAJ4QKFO8JYlJ3ChooG+oK813JxaR2JsZFcOWOS3aEoy6WZiTjjoz5WNurodvH8ewcpzEkhJy3exujUWKcJYQQKc5yc6Oxlx5ETdocyYp09LtbtOc7Suak6qUoACQsTluWls6mikZOn++9T/e7Dw7R29vKAXh0oH9NPghG4dlYSEQ4J6rLR+r0NnO7t03JRACrKT6enz83a8nq6evtYsfkgV2clcfGUiXaHpsY4TQgjEBcdweXTJgV189Piklqc8VFcmplodyhqkPyMCUxJjGFlSS1/2lZD06luvTpQfqEJYYQW56RQ1djBoaaO4VcOMG1dvWysaOSWeek4wgJ2kNmQJSIU5afxflUzz2yopGDqRK6Yrolb+Z4mhBEqtHotB+NVwtry4/T0uSnKD+5ms2NZUX46fW7D8bZuHrg+S+f7VX6hCWGEJifGkO2MC8qEUFxSS8bEccyfnGB3KOocsp1xzE6NIy9jAgtnaadB5R86eM0oLM5J4ZebqznZ2cuEmOAYNrqlo4d3K5u459rp+q0zgIkIv737chwi+v+k/EavEEahcI6TPrdh4/7gGezujbI6+tyGojxtXRToksZHMVHnO1B+pAlhFOZnJJA0PvKsXqWBrLiklhnJseSkxdkdilIqwGhCGIWwMGFRdgobKxo8munKbsfbuvjwYAtF+elahlBKnUUTwigtznHS3uXiowuYIN0ur5fWYQws03KRUmoImhBG6ZqZSUSGhwXFpDnFpbXMSYsnK0VHy1RKnU0TwijFRoVz5Yz+XsvGBO5gdzUtnew80qpDVSilzkkTghcsznFyuLmTqsZTdodyTqtK6wBYpjOjKaXOQROCFxRak9MHcmuj4pJaLp6SoFMvKqXOSROCF6RNGEduenzAjn5a2XCKPXVt2vdAKXVemhC8ZHGOkx1HTtDS0WN3KGdZVVqLCNyi5SKl1HloQvCSwpwU3AY27AusspExhuKSWi6flogzPtrucJRSAUwTgpfMTZ+AMz6K9fsCq2y0t66dqsYObV2klBqWJgQvCQsTrp/tZFNFI92uPrvD+Zvi0lrCw4Slc7VcpJQ6P00IXlSYk0JHTx8fVgdGr+Uz5aKrZyaRqIOkKaWGoQnBi67KSiI6Ioz1ATJHwq6aVo6eOK2ti5RSHtGE4EXREQ6uzkpm3d6GgOi1XFxSR2R4GDfkOu0ORSkVBDQheFlhTgrHWk+zr77d1jj63IZVpbUsyk4mPjo4Ju9RStlLE4KXXT+7v9ey3WWjrQdbaGjv1tZFSimPaULwspT4aPInJ9g+jEVxaS0xkY6/JSillBqOJgQfKJydwq6aVhrau2w5fm+fmzd211GY4yQmUqfNVkp5RhOCDyzO6b+Ja1ev5fcqmzjR2avlIqXUBdGE4AM5aXFclDDOtrJRcUkdcdHhXDsryZbjK6WCk0cJQUSWiEiFiFSKyMNDvH+XiDSKyC7rcbe1fL6IfCAi5SJSKiKfHbDNb0Tk4IBt5nvvx7KXiLA4J4V3DjTS1evfXstdvX2sLa9nSW4qUeEOvx5bKRXchk0IIuIAngGWAnOAO0RkzhCr/tEYM996PGct6wTuNMbkAkuAn4pIwoBtvjFgm12j+1ECy+IcJ129bt6vavLrcTftb6S926XlIqXUBfPkCuEyoNIYU22M6QFeBm71ZOfGmP3GmAPW81qgAUgeabDB5IrpicRGOvxeNiouqSUxNpIrZ0zy63GVUsHPk4RwEVAz4PVRa9lgn7TKQq+IyOTBb4rIZUAkUDVg8XetbX4iIlEXEnigiwp3cO2sZNb7ca7lzh4X6/c2cPO8VMIdentIKXVhPPnUkCGWDf6EKwYyjTF5wDrghY/tQCQNeAn4sjHGbS1+BJgNXAokAt8c8uAi94jINhHZ1tjY6EG4gWNxjpPjbd2UHWvzy/HW7W3gdG+fjl2klBoRTxLCUWDgN/4MoHbgCsaYZmNMt/XyWWDBmfdEJB54HXjUGLNlwDZ1pl838Gv6S1NnMcasMMYUGGMKkpODq9q0KDsZEVjnp17LxSW1OOOjuDQz0S/HU0qNLZ4khI+AmSIyTUQigduBlQNXsK4AzlgO7LWWRwJ/BV40xvz3UNuIiAC3AWUj/SEC1aTxUSyYMtEvk+acPN3LpopGluWlExY21EWdUkqd37AJwRjjAh4E1tD/Qf8nY0y5iDwpIsut1R6ympaWAA8Bd1nLPwNcC9w1RPPS34nIbmA3kAR8x2s/VQBZnOOk7FgbdSdP+/Q4a8vr6elza+sipdSISSAM0+ypgoICs23bNrvDuCAHjrdzw082853b5vKFK6b67Dh3Pr+Vg02n2PyNRfRfdCmlVD8R2W6MKRhuPW2K4mNZKeOZkhjj09FPm091815lE0V56ZoMlFIjpgnBx0SEwhwn71U109nj8skx3iirp89ttFyklBoVTQh+UJiTQo/LzTsHfNNrubiklqyU8cxOjfPJ/pVSoUETgh9cOi2RuOhwn5SN6k92sfVQi5aLlFKjpgnBDyIcYSzMTuHtfQ243d69if/67jqMgWX5acOvrJRS56EJwU8Kc1JoOtXDrqOtXt1vcUktuenxzEge79X9KqVCjyYEP1k4KwVHmHi1bFTT0smumla9mayU8gpNCH4yISaCgqkTWe/F0U+LS/tHELllnpaLlFKjpwnBj26Y42RffTs1LZ1e2V9xSR2XTElgcmKMV/anlAptmhD86Mxcy94oG1U2tLO3rk3LRUopr9GE4EfTkmKZnhzL+n2jLxsVl9QhouUipZT3aELwsxtynGypbqa9q3fE+zDGUFxayxXTJpESH+3F6JRSoUwTgp8tznHS22fYvH/kvZb31LVR3dih5SKllFdpQvCzS6YkkBATMar7CMUldYSHCUvmpnoxMqVUqNOE4GfhjjAWZaewoaIBV597+A0GMcZQXFLL1TOTSIyN9EGESqlQpQnBBoU5Tk509rLjyIX3Wt5Z08qx1tM6b7JSyus0Idjg2llJRDhG1mu5uKSWyPAwbsh1+iAypVQo04Rgg7joCC6fNol1F5gQ+tyG10vrWJSdTHx0hI+iU0qFKk0INinMSaGqsYODTR0eb7P1YAsN7d3aukgp5ROaEGwykl7LxaW1xEQ6uH52iq/CUkqFME0INpmcGEO2M87jslFvn5s3dtdRmOMkJjLcx9EppUKRJgQbLc5J4aNDJzjZOXyv5fcqmzjR2avlIqWUz2hCsFHhHCd9bsPG/cOPbVRcUkdcdDjXzkryQ2RKqVCkCcFG8zMSSBofybph5kjo6u1jbXk9S3JTiQp3+Ck6pVSo0YRgo7AwYVF2ChsrGug9T6/lTfsbae92ablIKeVTmhBsVjjHSXuXi48OtZxzneKSWhJjI7lyxiQ/RqaUCjWaEGx2zcwkIsPDWLdn6LJRZ4+L9XsbuHleKuEO/e9SSvmOfsLYLCYynCtnTGL9vuMYY856f93eBk739unYRUopn9OEEAAW5zg53NxJVeOps94rLqnFGR/FpZmJNkSmlAolmhACQGFOf8/jtwaVjU6e7mVTRSPL8tIJCxM7QlNKhRBNCAEgbcI4ctPjzxrGYm15PT19bm1dpJTyC00IAWJxjpMdR07QfKr7b8uKS+uYnDiO/IwJNkamlAoVHiUEEVkiIhUiUikiDw/x/l0i0igiu6zH3dby+SLygYiUi0ipiHx2wDbTRORDETkgIn8UkZCe/uuGHCduAxsqGgFoPtXNe5VNFOWlI6LlIqWU7w2bEETEATwDLAXmAHeIyJwhVv2jMWa+9XjOWtYJ3GmMyQWWAD8VkQTrve8DPzHGzAROAF8Z5c8S1OZeFI8zPupvZaPVZfX0uY2Wi5RSfuPJFcJlQKUxptoY0wO8DNzqyc6NMfuNMQes57VAA5As/V95rwdesVZ9AbjtQoMfS0SE62c72by/kW5XH8UltWSljGd2apzdoSmlQoQnCeEioGbA66PWssE+aZWFXhGRyYPfFJHLgEigCpgEtBpjXMPsM6QU5qTQ0dPHaztr+ehQi5aLlFJ+5UlCGOoTaXAPqmIg0xiTB6yj/xv//+xAJA14CfiyMcbt4T7PbHuPiGwTkW2NjY0ehBu8rspKIjoijO+u3osxsCw/ze6QlFIhxJOEcBQY+I0/A6gduIIxptkYc6Z5zLPAgjPviUg88DrwqDFmi7W4CUgQkTMzvZy1zwH7XmGMKTDGFCQnJ3sQbvCKjnBwdVYyJ0/3kpsez4zk8XaHpJQKIZ4khI+AmVaroEjgdmDlwBWsK4AzlgN7reWRwF+BF40x/31mBdM/RsMG4FPWoi8Br430hxhLznRS05vJSil/G3YuRmOMS0QeBNYADuB5Y0y5iDwJbDPGrAQeEpHlgAtoAe6yNv8McC0wSUTOLLvLGLML+Cbwsoh8B9gJ/Mp7P1bwWpafzv7jp/hswVm3YZRSyqdkqAHVAlVBQYHZtm2b3WEopVRQEZHtxpiC4dbTnspKKaUATQhKKaUsmhCUUkoBmhCUUkpZNCEopZQCNCEopZSyaEJQSikFaEJQSillCaqOaSLSCBwe4eZJ9I+hFCyCKV6N1XeCKd5gihWCK97RxjrVGDPsYHBBlRBGQ0S2edJTL1AEU7waq+8EU7zBFCsEV7z+ilVLRkoppQBNCEoppSyhlBBW2B3ABQqmeDVW3wmmeIMpVgiueP0Sa8jcQ1BKKXV+oXSFoJRS6jxCIiGIyCER2S0iu0QkoCZUEJHnRaRBRMoGLEsUkbdE5ID170Q7YxzoHPF+W0SOWed3l4jcbGeMZ4jIZBHZICJ7RaRcRL5mLQ+483ueWAP13EaLyFYRKbHifcJaPk1EPrTO7R+tWRMDNdbfiMjBAed2vt2xDiQiDhHZKSKrrNc+P7chkRAsi4wx8wOwmdlvgCWDlj0MrDfGzATWW68DxW84O16An1jnd74xZrWfYzoXF/B/jTE5wBXAAyIyh8A8v+eKFQLz3HYD1xtj8oH5wBIRuQL4Pv3xzgROAF+xMcYzzhUrwDcGnNtd9oU4pK9hTUds8fm5DaWEEJCMMZvpn3Z0oFuBF6znLwC3+TWo8zhHvAHJGFNnjNlhPW+n/4/rIgLw/J4n1oBk+p2yXkZYDwNcD7xiLQ+Uc3uuWAOWiGQAtwDPWa8FP5zbUEkIBlgrIttF5B67g/GA0xhTB/0fFECKzfF44kERKbVKSraXYAYTkUzgYuBDAvz8DooVAvTcWiWNXUAD8BZQBbQaY1zWKkcJkKQ2OFZjzJlz+13r3P5ERKJsDHGwnwL/DLit15Pww7kNlYRwlTHmEmAp/Zfi19od0BjzC2AG/ZfjdcCP7A3n40RkPPBn4OvGmDa74zmfIWIN2HNrjOkzxswHMoDLgJyhVvNvVEMbHKuIzAUeAWYDlwKJwDdtDPFvRGQZ0GCM2T5w8RCrev3chkRCMMbUWv82AH+l/5c3kB0XkTQA698Gm+M5L2PMcesPzg08SwCdXxGJoP8D9nfGmL9YiwPy/A4VayCf2zOMMa3ARvrvfSSISLj1VgZQa1dcQxkQ6xKrTGeMMd3Arwmcc3sVsFxEDgEv018q+il+OLdjPiGISKyIxJ15DtwIlJ1/K9utBL5kPf8S8JqNsQzrzIer5X8RIOfXqrv+CthrjPnxgLcC7vyeK9YAPrfJIpJgPR8HFNJ/32MD8ClrtUA5t0PFum/AlwKhvx4fEOfWGPOIMSbDGJMJ3A68bYz5PH44t2O+Y5qITKf/qgAgHPi9Mea7Nob0MSLyB2Ah/aMZHge+BbwK/AmYAhwBPm2MCYgbueeIdyH9JQ0DHALuPVOjt5OIXA28A+zmf2qx/0J/bT6gzu95Yr2DwDy3efTf2HTQ/8XyT8aYJ62/t5fpL8HsBL5gfQO3zXlifRtIpr8cswv46oCbzwFBRBYC/2SMWeaPczvmE4JSSinPjPmSkVJKKc9oQlBKKQVoQlBKKWXRhKCUUgrQhKCUUsqiCUEppRSgCUEppZRFE4JSSikA/j+7TUS0GC6pKAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "n_scores = []\n",
    "estimator_nums = [5, 10, 15, 20, 25, 30, 35, 40]\n",
    "for estimator_num in estimator_nums:\n",
    "    clf = AdaBoostClassifier(n_estimators=estimator_num, base_estimator=dtc)\n",
    "    n_score = cross_val_score(clf, X_train, y_train, cv=5)\n",
    "    n_scores.append(np.mean(n_score))\n",
    "plt.plot(estimator_nums, n_scores)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "当 estimators=20 的时候，score最高，大概有0.544，虽然跟单个决策树的 score 的值相差不大，但总体还是有所提升。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "abc = AdaBoostClassifier(n_estimators=20)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "DecisionTreeClassifier(class_weight=None, criterion='gini', max_depth=None,\n",
       "            max_features=0.2, max_leaf_nodes=None,\n",
       "            min_impurity_decrease=0.0, min_impurity_split=None,\n",
       "            min_samples_leaf=1, min_samples_split=2,\n",
       "            min_weight_fraction_leaf=0.0, presort=False, random_state=None,\n",
       "            splitter='best')"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "abc.fit(X_train, y_train)\n",
    "dtc.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_dtc = dtc.predict(X_test)\n",
    "y_abc = abc.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "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>学历要求</th>\n",
       "      <th>工作经验</th>\n",
       "      <th>薪资(单个决策树)</th>\n",
       "      <th>薪资(boosting)</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>本科</td>\n",
       "      <td>1-3年</td>\n",
       "      <td>10k-20k</td>\n",
       "      <td>10k-20k</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>本科</td>\n",
       "      <td>3-5年</td>\n",
       "      <td>20k-30k</td>\n",
       "      <td>20k-30k</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>本科</td>\n",
       "      <td>5-10年</td>\n",
       "      <td>30k-40k</td>\n",
       "      <td>20k-30k</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>本科</td>\n",
       "      <td>应届毕业生</td>\n",
       "      <td>10k-20k</td>\n",
       "      <td>10k-20k</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>本科</td>\n",
       "      <td>不限</td>\n",
       "      <td>10k-20k</td>\n",
       "      <td>10k-20k</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>本科</td>\n",
       "      <td>1年以下</td>\n",
       "      <td>10k-20k</td>\n",
       "      <td>10k-20k</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>大专</td>\n",
       "      <td>1-3年</td>\n",
       "      <td>5k-10k</td>\n",
       "      <td>5k-10k</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>大专</td>\n",
       "      <td>3-5年</td>\n",
       "      <td>10k-20k</td>\n",
       "      <td>10k-20k</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>大专</td>\n",
       "      <td>5-10年</td>\n",
       "      <td>30k-40k</td>\n",
       "      <td>10k-20k</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>大专</td>\n",
       "      <td>应届毕业生</td>\n",
       "      <td>10k-20k</td>\n",
       "      <td>5k-10k</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>大专</td>\n",
       "      <td>不限</td>\n",
       "      <td>5k-10k</td>\n",
       "      <td>5k-10k</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>大专</td>\n",
       "      <td>1年以下</td>\n",
       "      <td>10k-20k</td>\n",
       "      <td>5k-10k</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>不限</td>\n",
       "      <td>1-3年</td>\n",
       "      <td>10k-20k</td>\n",
       "      <td>10k-20k</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>不限</td>\n",
       "      <td>3-5年</td>\n",
       "      <td>20k-30k</td>\n",
       "      <td>20k-30k</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>不限</td>\n",
       "      <td>5-10年</td>\n",
       "      <td>30k-40k</td>\n",
       "      <td>20k-30k</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>不限</td>\n",
       "      <td>应届毕业生</td>\n",
       "      <td>10k-20k</td>\n",
       "      <td>10k-20k</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>不限</td>\n",
       "      <td>不限</td>\n",
       "      <td>20k-30k</td>\n",
       "      <td>10k-20k</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>不限</td>\n",
       "      <td>1年以下</td>\n",
       "      <td>10k-20k</td>\n",
       "      <td>10k-20k</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>硕士</td>\n",
       "      <td>1-3年</td>\n",
       "      <td>20k-30k</td>\n",
       "      <td>20k-30k</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>硕士</td>\n",
       "      <td>3-5年</td>\n",
       "      <td>20k-30k</td>\n",
       "      <td>20k-30k</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>硕士</td>\n",
       "      <td>5-10年</td>\n",
       "      <td>30k-40k</td>\n",
       "      <td>20k-30k</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>硕士</td>\n",
       "      <td>应届毕业生</td>\n",
       "      <td>10k-20k</td>\n",
       "      <td>20k-30k</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>硕士</td>\n",
       "      <td>不限</td>\n",
       "      <td>10k-20k</td>\n",
       "      <td>20k-30k</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>硕士</td>\n",
       "      <td>1年以下</td>\n",
       "      <td>10k-20k</td>\n",
       "      <td>10k-20k</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   学历要求   工作经验 薪资(单个决策树) 薪资(boosting)\n",
       "0    本科   1-3年   10k-20k      10k-20k\n",
       "1    本科   3-5年   20k-30k      20k-30k\n",
       "2    本科  5-10年   30k-40k      20k-30k\n",
       "3    本科  应届毕业生   10k-20k      10k-20k\n",
       "4    本科     不限   10k-20k      10k-20k\n",
       "5    本科   1年以下   10k-20k      10k-20k\n",
       "6    大专   1-3年    5k-10k       5k-10k\n",
       "7    大专   3-5年   10k-20k      10k-20k\n",
       "8    大专  5-10年   30k-40k      10k-20k\n",
       "9    大专  应届毕业生   10k-20k       5k-10k\n",
       "10   大专     不限    5k-10k       5k-10k\n",
       "11   大专   1年以下   10k-20k       5k-10k\n",
       "12   不限   1-3年   10k-20k      10k-20k\n",
       "13   不限   3-5年   20k-30k      20k-30k\n",
       "14   不限  5-10年   30k-40k      20k-30k\n",
       "15   不限  应届毕业生   10k-20k      10k-20k\n",
       "16   不限     不限   20k-30k      10k-20k\n",
       "17   不限   1年以下   10k-20k      10k-20k\n",
       "18   硕士   1-3年   20k-30k      20k-30k\n",
       "19   硕士   3-5年   20k-30k      20k-30k\n",
       "20   硕士  5-10年   30k-40k      20k-30k\n",
       "21   硕士  应届毕业生   10k-20k      20k-30k\n",
       "22   硕士     不限   10k-20k      20k-30k\n",
       "23   硕士   1年以下   10k-20k      10k-20k"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_data['薪资(单个决策树)'] = y_dtc\n",
    "test_data['薪资(boosting)'] = y_abc\n",
    "test_data"
   ]
  }
 ],
 "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": "3.6.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}