客户逾期分析/test.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "<a id='1'></a>\n",
    "# 数据探索"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(4754, 90)\n"
     ]
    }
   ],
   "source": [
    "#encoding: utf-8\n",
    "\n",
    "import pandas as pd\n",
    "data = pd.read_csv('data/data.csv', encoding='GB18030')\n",
    "print(data.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 去重"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "去掉重复数据，即那些属性都一样的数据。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(4754, 90)\n"
     ]
    }
   ],
   "source": [
    "data.drop_duplicates(inplace=True)\n",
    "print(data.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<a id='1.1'></a>\n",
    "## 删除和预测无关的数据"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**删除原则**\n",
    "\n",
    "* 特征只有单一的值，即该特征所有样本值都一样\n",
    "* 特征属性值个数和行数相同\n",
    "* 特征和预测无关"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 240,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(4754, 0)\n",
      "(4754, 83)\n",
      "(4754, 7)\n",
      "(4754, 0)\n",
      "(4754, 0)\n",
      "(4754, 0)\n",
      "<class 'pandas.core.series.Series'>\n",
      "student_feature                     0.630627\n",
      "cross_consume_count_last_1_month    0.089609\n",
      "query_org_count                     0.063946\n",
      "query_cash_count                    0.063946\n",
      "latest_six_month_apply              0.063946\n",
      "latest_three_month_apply            0.063946\n",
      "query_sum_count                     0.063946\n",
      "latest_one_month_apply              0.063946\n",
      "query_finance_count                 0.063946\n",
      "apply_credibility                   0.063946\n",
      "dtype: float64\n",
      "0       1\n",
      "1       0\n",
      "2       1\n",
      "3       0\n",
      "4       1\n",
      "5       0\n",
      "6       0\n",
      "7       0\n",
      "8       0\n",
      "9       1\n",
      "10      1\n",
      "11      0\n",
      "12      0\n",
      "13      0\n",
      "14      1\n",
      "15      1\n",
      "16      1\n",
      "17      0\n",
      "18      0\n",
      "19      1\n",
      "20      1\n",
      "21      0\n",
      "22      0\n",
      "23      0\n",
      "24      0\n",
      "25      1\n",
      "26      0\n",
      "27      0\n",
      "28      0\n",
      "29      0\n",
      "       ..\n",
      "4724    1\n",
      "4725    0\n",
      "4726    0\n",
      "4727    0\n",
      "4728    0\n",
      "4729    0\n",
      "4730    1\n",
      "4731    0\n",
      "4732    0\n",
      "4733    1\n",
      "4734    1\n",
      "4735    1\n",
      "4736    0\n",
      "4737    1\n",
      "4738    1\n",
      "4739    0\n",
      "4740    0\n",
      "4741    0\n",
      "4742    0\n",
      "4743    0\n",
      "4744    1\n",
      "4745    0\n",
      "4746    0\n",
      "4747    0\n",
      "4748    0\n",
      "4749    1\n",
      "4750    0\n",
      "4751    0\n",
      "4752    0\n",
      "4753    0\n",
      "Name: status, Length: 4754, dtype: int64\n"
     ]
    },
    {
     "ename": "AttributeError",
     "evalue": "'numpy.ndarray' object has no attribute 'select_dtypes'",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mAttributeError\u001b[0m                            Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-240-82745afcdbb8>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m     73\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     74\u001b[0m \u001b[0;31m# 字符型变量，观察缺失值以及分布\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 75\u001b[0;31m \u001b[0mX_str\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mX\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mselect_dtypes\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mexclude\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m'number'\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcopy\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     76\u001b[0m \u001b[0mX_str\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdescribe\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     77\u001b[0m \u001b[0;31m# 众数填充\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mAttributeError\u001b[0m: 'numpy.ndarray' object has no attribute 'select_dtypes'"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAnAAAAE9CAYAAACLPV+MAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<Figure size 720x360 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAn0AAAE9CAYAAAB+w9Q8AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<Figure size 720x360 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAmIAAAE9CAYAAACoZg5ZAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nQmCC\n",
      "text/plain": [
       "<Figure size 720x360 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAmIAAAE9CAYAAACoZg5ZAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nQUySJKmQ/wAFYsEWMmI1QAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 720x360 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAnEAAAE9CAYAAABk/zSyAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<Figure size 720x360 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# 查看每列的类型\n",
    "types = data.dtypes\n",
    "# print(len(types))\n",
    "# print types[0:45]\n",
    "# print('-----------------------------------------------------------')\n",
    "# print types[45:]\n",
    "\n",
    "print data.select_dtypes(include=['bool']).shape\n",
    "print data.select_dtypes(include=['number']).shape\n",
    "print data.select_dtypes(include=['object']).shape\n",
    "print data.select_dtypes(include=['datetime']).shape\n",
    "print data.select_dtypes(include=['timedelta']).shape\n",
    "print data.select_dtypes(include=['category']).shape\n",
    "\n",
    "# 特征类型划分：数字型特征、字符型特征、日期特征\n",
    "# fancyimpute是python的第三方工具包，主要提供了各种矩阵计算、填充算法的实现。\n",
    "# https://blog.csdn.net/jingyi130705008/article/details/82796283\n",
    "\n",
    "# 数据可视化：https://www.cnblogs.com/jin-liang/p/9011771.html\n",
    "\n",
    "# 取出数值型特征，观察缺失及分布情况\n",
    "# 取出数值型变量\n",
    "X_num = data.select_dtypes(['number']).copy()\n",
    "# 查看缺失情况\n",
    "num_miss_rate = 1 - X_num.count() / len(X_num)\n",
    "print type(num_miss_rate)\n",
    "num_miss_rate.sort_values(ascending=False, inplace=True)\n",
    "print(num_miss_rate[:10])\n",
    "num_miss_rate.plot()\n",
    "\n",
    "X_num = data.select_dtypes(['number']).copy()\n",
    "# 观察数据分布\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "%matplotlib inline\n",
    "plt.rcParams['font.sans-serif'] = ['SimHei']  # 用来正常显示中文标签\n",
    "plt.rcParams['axes.unicode_minus'] = False  # 用来正常显示负号\n",
    " \n",
    "print y\n",
    "\n",
    "for i, col in enumerate(X_num.columns[:5]):\n",
    "    plt.figure(i + 1, figsize=(10, 5))\n",
    "    # # 密度分布图\n",
    "    # sns.distplot(X_num[col][y==1].dropna(), hist=True, kde=True)\n",
    "    # sns.distplot(X_num[col][y==0].dropna(), hist=True, kde=True)\n",
    "    # 分箱图\n",
    "    # sns.boxplot(y=X_num[col], x=y)\n",
    "    # 小提琴图\n",
    "    # sns.violinplot(y=X_num[col])\n",
    "    sns.violinplot(x=y, y=X_num[col])\n",
    "\n",
    "# 从图像可以看出，数据中含有极端值的比较多，可使用 IQR 方法对极端值进行处理。\n",
    "# 也就是说，分别使用 Q3 + 1.5IQR和 Q1 - 1.5IQR 对上下极端值进行修正。\n",
    "# 极端值处理函数\n",
    "def iqr_outlier(x, thre=1.5):\n",
    "    x_cl = x.copy()\n",
    "    q25, q75 = x.quantile(q=[0.25, 0.75])\n",
    "    iqr = q75 - q25\n",
    "    top = q75 + thre * iqr\n",
    "    bottom = q25 - thre * iqr\n",
    " \n",
    "    x_cl[x_cl > top] = top\n",
    "    x_cl[x_cl < bottom] = bottom\n",
    " \n",
    "    return x_cl\n",
    " \n",
    "# 对进行缺失值填充后的数据集进行极端值修正\n",
    "X_num_cl = pd.DataFrame()\n",
    "for col in X_num.columns:\n",
    "    X_num_cl[col] = iqr_outlier(X_num[col])\n",
    "\n",
    "X_num = X_num_cl\n",
    "\n",
    "# 字符型变量，观察缺失值以及分布\n",
    "X_str = X.select_dtypes(exclude='number').copy()\n",
    "X_str.describe()\n",
    "# 众数填充\n",
    "X_str['reg_preference_for_trad'] = X_str['reg_preference_for_trad'].fillna(X_str['reg_preference_for_trad'].mode()[0])\n",
    "# 哑变量\n",
    "X_str_dummy = pd.get_dummies(X_str['reg_preference_for_trad'])\n",
    "# 数字编码\n",
    "X_str_map = X_str['reg_preference_for_trad'].map({'一线城市': 0, '二线城市': 1,'三线城市': 2,  '其他城市': 3, '境外': 4})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Unnamed: 0\n",
      "custid\n",
      "trade_no\n",
      "bank_card_no\n",
      "source\n"
     ]
    }
   ],
   "source": [
    "# 1表示某列无缺失值情况下，所有值都一样；\n",
    "# 2表示有缺失值情况下，所有值都一样\n",
    "# count表示无缺失值情况下，所有值都不一样\n",
    "# 表示有缺失值情况下，所有值都不一样\n",
    "# 但对于离散值，其实所有值都一样也没关系\n",
    "for i in data.columns:\n",
    "    count = data[i].count()\n",
    "    if len(list(data[i].unique())) in [1, count, count - 1]:\n",
    "        print(i)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# 'source'和'bank_card_no'值无区分度\n",
    "# ‘custid’、‘trade_no’、‘id_name’和预测无关\n",
    "data.drop(['custid', 'trade_no', 'bank_card_no', 'id_name', 'source'], axis = 1, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# 'Unnamed: 0'和预测无关\n",
    "data.drop(['Unnamed: 0'], axis = 1, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "first_transaction_time = data['first_transaction_time']\n",
    "latest_query_time = data['latest_query_time']\n",
    "loans_latest_time = data['loans_latest_time']\n",
    "# 先删除data\n",
    "data.drop(['first_transaction_time', 'latest_query_time', 'loans_latest_time'], axis = 1, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1694\n",
      "208\n",
      "233\n"
     ]
    }
   ],
   "source": [
    "# 探索处理日期类型的方法\n",
    "print len(first_transaction_time.unique())\n",
    "print len(latest_query_time.unique())\n",
    "print len(loans_latest_time.unique())\n",
    "\n",
    "# print first_transaction_time[1]\n",
    "\n",
    "latest_query_time_month = pd.to_datetime(latest_query_time).dt.month # 月\n",
    "latest_query_time_weekday = pd.to_datetime(latest_query_time).dt.weekday # 星期\n",
    "latest_query_time_day = pd.to_datetime(latest_query_time).dt.day # 天\n",
    " \n",
    "loans_latest_time_month = pd.to_datetime(loans_latest_time).dt.month # 月\n",
    "loans_latest_time_weekday = pd.to_datetime(loans_latest_time).dt.weekday # 星期\n",
    "loans_latest_time_day = pd.to_datetime(loans_latest_time).dt.day # 天\n",
    "\n",
    "data['latest_query_time_month'] = latest_query_time_month\n",
    "data['latest_query_time_weekday'] = latest_query_time_weekday\n",
    "data['latest_query_time_day'] = latest_query_time_day\n",
    "data['loans_latest_time_month'] = loans_latest_time_month\n",
    "data['loans_latest_time_weekday'] = loans_latest_time_weekday\n",
    "data['loans_latest_time_day'] = loans_latest_time_day\n",
    "\n",
    "latest_query_time_month = latest_query_time_month.as_matrix()\n",
    "latest_query_time_month = latest_query_time_month.reshape(len(latest_query_time_month), 1)\n",
    "\n",
    "latest_query_time_weekday = latest_query_time_weekday.as_matrix()\n",
    "latest_query_time_weekday = latest_query_time_weekday.reshape(len(latest_query_time_weekday), 1)\n",
    "\n",
    "latest_query_time_day = latest_query_time_day.as_matrix()\n",
    "latest_query_time_day = latest_query_time_day.reshape(len(latest_query_time_day), 1)\n",
    "\n",
    "loans_latest_time_month = loans_latest_time_month.as_matrix()\n",
    "loans_latest_time_month = loans_latest_time_month.reshape(len(loans_latest_time_month), 1)\n",
    "\n",
    "loans_latest_time_weekday = loans_latest_time_weekday.as_matrix()\n",
    "loans_latest_time_weekday = loans_latest_time_weekday.reshape(len(loans_latest_time_weekday), 1)\n",
    "\n",
    "loans_latest_time_day = loans_latest_time_day.as_matrix()\n",
    "loans_latest_time_day = loans_latest_time_day.reshape(len(loans_latest_time_day), 1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<a id='1.2'></a>\n",
    "## 查看行列缺失比例"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.04% low_volume_percent\n",
      "0.04% middle_volume_percent\n",
      "0.00% take_amount_in_later_12_month_highest\n",
      "0.06% trans_amount_increase_rate_lately\n",
      "0.04% trans_activity_month\n",
      "0.04% trans_activity_day\n",
      "0.04% transd_mcc\n",
      "0.17% trans_days_interval_filter\n",
      "0.04% trans_days_interval\n",
      "0.04% regional_mobility\n",
      "63.06% student_feature\n",
      "0.00% repayment_capability\n",
      "0.00% is_high_user\n",
      "0.04% number_of_trans_from_2011\n",
      "0.00% historical_trans_amount\n",
      "0.04% historical_trans_day\n",
      "0.04% rank_trad_1_month\n",
      "0.00% trans_amount_3_month\n",
      "0.04% avg_consume_less_12_valid_month\n",
      "0.00% abs\n",
      "0.04% top_trans_count_last_1_month\n",
      "0.00% avg_price_last_12_month\n",
      "2.19% avg_price_top_last_12_valid_month\n",
      "0.04% reg_preference_for_trad\n",
      "0.17% trans_top_time_last_1_month\n",
      "0.17% trans_top_time_last_6_month\n",
      "0.17% consume_top_time_last_1_month\n",
      "0.17% consume_top_time_last_6_month\n",
      "8.96% cross_consume_count_last_1_month\n",
      "0.34% trans_fail_top_count_enum_last_1_month\n",
      "0.34% trans_fail_top_count_enum_last_6_month\n",
      "0.34% trans_fail_top_count_enum_last_12_month\n",
      "0.55% consume_mini_time_last_1_month\n",
      "0.00% max_cumulative_consume_later_1_month\n",
      "0.17% max_consume_count_later_6_month\n",
      "0.25% railway_consume_count_last_12_month\n",
      "0.00% pawns_auctions_trusts_consume_last_1_month\n",
      "0.00% pawns_auctions_trusts_consume_last_6_month\n",
      "0.25% jewelry_consume_count_last_6_month\n",
      "0.00% status\n",
      "0.04% first_transaction_day\n",
      "0.04% trans_day_last_12_month\n",
      "6.39% apply_score\n",
      "6.39% apply_credibility\n",
      "6.39% query_org_count\n",
      "6.39% query_finance_count\n",
      "6.39% query_cash_count\n",
      "6.39% query_sum_count\n",
      "6.39% latest_one_month_apply\n",
      "6.39% latest_three_month_apply\n",
      "6.39% latest_six_month_apply\n",
      "6.25% loans_score\n",
      "6.25% loans_credibility_behavior\n",
      "6.25% loans_count\n",
      "6.25% loans_settle_count\n",
      "6.25% loans_overdue_count\n",
      "6.25% loans_org_count_behavior\n",
      "6.25% consfin_org_count_behavior\n",
      "6.25% loans_cash_count\n",
      "6.25% latest_one_month_loan\n",
      "6.25% latest_three_month_loan\n",
      "6.25% latest_six_month_loan\n",
      "6.25% history_suc_fee\n",
      "6.25% history_fail_fee\n",
      "6.25% latest_one_month_suc\n",
      "6.25% latest_one_month_fail\n",
      "6.25% loans_long_time\n",
      "6.25% loans_credit_limit\n",
      "6.25% loans_credibility_limit\n",
      "6.25% loans_org_count_current\n",
      "6.25% loans_product_count\n",
      "6.25% loans_max_limit\n",
      "6.25% loans_avg_limit\n",
      "6.25% consfin_credit_limit\n",
      "6.25% consfin_credibility\n",
      "6.25% consfin_org_count_current\n",
      "6.25% consfin_product_count\n",
      "6.25% consfin_max_limit\n",
      "6.25% consfin_avg_limit\n",
      "6.39% latest_query_day\n",
      "6.25% loans_latest_day\n",
      "6.39% latest_query_time_month\n",
      "6.39% latest_query_time_weekday\n",
      "6.39% latest_query_time_day\n",
      "6.25% loans_latest_time_month\n",
      "6.25% loans_latest_time_weekday\n",
      "6.25% loans_latest_time_day\n"
     ]
    }
   ],
   "source": [
    "# 统计各个列缺失值所占比例\n",
    "for i in data.columns:\n",
    "    d = len(data) - data[i].count()\n",
    "    r = (float(d) / len(data)) * 100\n",
    "    # rate = '%.2f%%' % r\n",
    "    # print 'name: ', str(i).ljust(10),'d: ', str(d).ljust(4), 'rate: ', rate\n",
    "    print '%.2f%%' % r, i\n",
    "\n",
    "# 由下图统计可以看出，‘student_feature’列缺失一半以上，且本列为类别类型，可以将缺失值用-1填充，相当于“是否缺失”当成另一种类别。\n",
    "# 其他列缺失概率比较小，可以用中值填充。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(4754, 87)\n",
      "(4754, 88)\n",
      "(4754, 89)\n"
     ]
    }
   ],
   "source": [
    "# 缺失个数作为一种特征，衡量用户的信息完善程度\n",
    "miss_rate = []\n",
    "miss_number = []\n",
    "for i in range(len(data)):\n",
    "    temp = float((data[i:i+1]).count().sum()) / len(data.columns)\n",
    "    miss_rate.append(temp)\n",
    "    miss_number.append(len(data.columns) - (data[i:i+1]).count().sum())\n",
    "\n",
    "print data.shape\n",
    "data['miss_rate'] = miss_rate\n",
    "print data.shape\n",
    "data['miss_number'] = miss_number\n",
    "print data.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0       0.988506\n",
      "1       1.000000\n",
      "2       0.988506\n",
      "3       0.988506\n",
      "4       0.988506\n",
      "5       1.000000\n",
      "6       0.965517\n",
      "7       0.471264\n",
      "8       0.988506\n",
      "9       1.000000\n",
      "10      0.988506\n",
      "11      1.000000\n",
      "12      1.000000\n",
      "13      0.977011\n",
      "14      0.988506\n",
      "15      0.988506\n",
      "16      0.988506\n",
      "17      1.000000\n",
      "18      1.000000\n",
      "19      0.988506\n",
      "20      1.000000\n",
      "21      0.988506\n",
      "22      0.988506\n",
      "23      0.988506\n",
      "24      1.000000\n",
      "25      0.977011\n",
      "26      1.000000\n",
      "27      1.000000\n",
      "28      1.000000\n",
      "29      0.850575\n",
      "          ...   \n",
      "4724    1.000000\n",
      "4725    0.977011\n",
      "4726    0.988506\n",
      "4727    0.988506\n",
      "4728    0.988506\n",
      "4729    0.977011\n",
      "4730    1.000000\n",
      "4731    1.000000\n",
      "4732    0.977011\n",
      "4733    0.988506\n",
      "4734    0.988506\n",
      "4735    1.000000\n",
      "4736    1.000000\n",
      "4737    1.000000\n",
      "4738    0.988506\n",
      "4739    0.988506\n",
      "4740    0.988506\n",
      "4741    1.000000\n",
      "4742    1.000000\n",
      "4743    0.988506\n",
      "4744    0.988506\n",
      "4745    0.988506\n",
      "4746    0.482759\n",
      "4747    0.988506\n",
      "4748    0.977011\n",
      "4749    1.000000\n",
      "4750    1.000000\n",
      "4751    0.988506\n",
      "4752    0.988506\n",
      "4753    1.000000\n",
      "Name: miss_rate, Length: 4754, dtype: float64\n"
     ]
    }
   ],
   "source": [
    "print data['miss_rate']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<a id='1.3'></a>\n",
    "## 处理类别特征"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "3.0    1950\n",
       "2.0    1515\n",
       "4.0     802\n",
       "1.0     446\n",
       "5.0      39\n",
       "Name: regional_mobility, dtype: int64"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 'regional_mobility'列的统计，按类别特征处理\n",
    "data['regional_mobility'].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "一线城市    3403\n",
       "三线城市    1064\n",
       "境外       150\n",
       "二线城市     131\n",
       "其他城市       4\n",
       "Name: reg_preference_for_trad, dtype: int64"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 'reg_preference_for_trad'列的统计，按类别特征处理\n",
    "data['reg_preference_for_trad'].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1.0    1754\n",
       "2.0       2\n",
       "Name: student_feature, dtype: int64"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 'student_feature'列的统计，按类别特征处理\n",
    "data['student_feature'].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0    4701\n",
       "1      53\n",
       "Name: is_high_user, dtype: int64"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 'is_high_user'列的统计，按类别特征处理\n",
    "data['is_high_user'].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0    3561\n",
       "1    1193\n",
       "Name: status, dtype: int64"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 'status'列的统计，预测变量，正负样本接近1：3，可以不做处理。\n",
    "data['status'].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(4754, 83)\n",
      "(4754, 89)\n"
     ]
    }
   ],
   "source": [
    "# 将刚刚被归类为类别变量和预测变量的列去掉，生成data_temp，数值特征为77维，类别特征为5维\n",
    "data_temp = data\n",
    "data_temp = data_temp.drop(['regional_mobility', 'reg_preference_for_trad', 'student_feature', 'is_high_user', 'status', 'miss_rate'], axis = 1)\n",
    "print data_temp.shape\n",
    "print data.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<a id='1.4'></a>\n",
    "## 特征标准差探索"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "83\n",
      "0.04 low_volume_percent\n",
      "0.14 middle_volume_percent\n",
      "3923.97 take_amount_in_later_12_month_highest\n",
      "694.18 trans_amount_increase_rate_lately\n",
      "0.20 trans_activity_month\n",
      "0.17 trans_activity_day\n",
      "4.48 transd_mcc\n",
      "22.72 trans_days_interval_filter\n",
      "16.47 trans_days_interval\n",
      "52217.83 repayment_capability\n",
      "10.06 number_of_trans_from_2011\n",
      "320493.12 historical_trans_amount\n",
      "99.69 historical_trans_day\n",
      "0.26 rank_trad_1_month\n",
      "101746.13 trans_amount_3_month\n",
      "1.39 avg_consume_less_12_valid_month\n",
      "27007.60 abs\n",
      "0.35 top_trans_count_last_1_month\n",
      "765.87 avg_price_last_12_month\n",
      "0.10 avg_price_top_last_12_valid_month\n",
      "5.32 trans_top_time_last_1_month\n",
      "12.96 trans_top_time_last_6_month\n",
      "5.46 consume_top_time_last_1_month\n",
      "13.13 consume_top_time_last_6_month\n",
      "2.34 cross_consume_count_last_1_month\n",
      "1.91 trans_fail_top_count_enum_last_1_month\n",
      "4.46 trans_fail_top_count_enum_last_6_month\n",
      "4.76 trans_fail_top_count_enum_last_12_month\n",
      "374267.23 consume_mini_time_last_1_month\n",
      "10813.45 max_cumulative_consume_later_1_month\n",
      "5.68 max_consume_count_later_6_month\n",
      "0.48 railway_consume_count_last_12_month\n",
      "6616.69 pawns_auctions_trusts_consume_last_1_month\n",
      "28191.13 pawns_auctions_trusts_consume_last_6_month\n",
      "0.20 jewelry_consume_count_last_6_month\n",
      "537.11 first_transaction_day\n",
      "19.07 trans_day_last_12_month\n",
      "51.17 apply_score\n",
      "4.17 apply_credibility\n",
      "7.04 query_org_count\n",
      "3.81 query_finance_count\n",
      "2.60 query_cash_count\n",
      "11.30 query_sum_count\n",
      "4.53 latest_one_month_apply\n",
      "7.62 latest_three_month_apply\n",
      "9.27 latest_six_month_apply\n",
      "60.95 loans_score\n",
      "2.23 loans_credibility_behavior\n",
      "24.61 loans_count\n",
      "21.69 loans_settle_count\n",
      "3.15 loans_overdue_count\n",
      "7.45 loans_org_count_behavior\n",
      "2.97 consfin_org_count_behavior\n",
      "5.37 loans_cash_count\n",
      "1.50 latest_one_month_loan\n",
      "3.46 latest_three_month_loan\n",
      "10.83 latest_six_month_loan\n",
      "30.35 history_suc_fee\n",
      "25.09 history_fail_fee\n",
      "1.94 latest_one_month_suc\n",
      "3.89 latest_one_month_fail\n",
      "35.77 loans_long_time\n",
      "708.95 loans_credit_limit\n",
      "10.85 loans_credibility_limit\n",
      "5.37 loans_org_count_current\n",
      "5.76 loans_product_count\n",
      "1474.21 loans_max_limit\n",
      "583.42 loans_avg_limit\n",
      "7371.26 consfin_credit_limit\n",
      "14.54 consfin_credibility\n",
      "2.97 consfin_org_count_current\n",
      "3.41 consfin_product_count\n",
      "14301.04 consfin_max_limit\n",
      "5679.42 consfin_avg_limit\n",
      "37.73 latest_query_day\n",
      "53.49 loans_latest_day\n",
      "1.38 latest_query_time_month\n",
      "1.95 latest_query_time_weekday\n",
      "9.39 latest_query_time_day\n",
      "3.08 loans_latest_time_month\n",
      "1.96 loans_latest_time_weekday\n",
      "9.46 loans_latest_time_day\n",
      "10.80 miss_number\n",
      "82\n"
     ]
    }
   ],
   "source": [
    "# 统计各个列标准差，将标准差小于0.1的特征剔除，数值特征变为71维\n",
    "print (len(data_temp.columns))\n",
    "for i in data_temp.columns:\n",
    "    r = data_temp[i].std()\n",
    "    print '%.2f' % r, i\n",
    "    \n",
    "    if r < 0.1:\n",
    "        data_temp = data_temp.drop([i], axis = 1)\n",
    "print (len(data_temp.columns))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "<a id='1.5'></a>\n",
    "## 缺失值填充和特征编码"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**缺失值填充**\n",
    "\n",
    "* 众数，一般针对类别变量\n",
    "* 中位数，一般针对连续变量\n",
    "* 分箱处理，将缺失值作为一类处理。对于一些没有值可以认为此值不存在的情况，不存在就属于一个类别。\n",
    "* 或者用工具填充，决策树等模型"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "middle_volume_percent\n",
      "trans_amount_increase_rate_lately\n",
      "trans_activity_month\n",
      "trans_activity_day\n",
      "transd_mcc\n",
      "trans_days_interval_filter\n",
      "trans_days_interval\n",
      "number_of_trans_from_2011\n",
      "historical_trans_day\n",
      "rank_trad_1_month\n",
      "avg_consume_less_12_valid_month\n",
      "top_trans_count_last_1_month\n",
      "avg_price_top_last_12_valid_month\n",
      "trans_top_time_last_1_month\n",
      "trans_top_time_last_6_month\n",
      "consume_top_time_last_1_month\n",
      "consume_top_time_last_6_month\n",
      "cross_consume_count_last_1_month\n",
      "trans_fail_top_count_enum_last_1_month\n",
      "trans_fail_top_count_enum_last_6_month\n",
      "trans_fail_top_count_enum_last_12_month\n",
      "consume_mini_time_last_1_month\n",
      "max_consume_count_later_6_month\n",
      "railway_consume_count_last_12_month\n",
      "jewelry_consume_count_last_6_month\n",
      "first_transaction_day\n",
      "trans_day_last_12_month\n",
      "apply_score\n",
      "apply_credibility\n",
      "query_org_count\n",
      "query_finance_count\n",
      "query_cash_count\n",
      "query_sum_count\n",
      "latest_one_month_apply\n",
      "latest_three_month_apply\n",
      "latest_six_month_apply\n",
      "loans_score\n",
      "loans_credibility_behavior\n",
      "loans_count\n",
      "loans_settle_count\n",
      "loans_overdue_count\n",
      "loans_org_count_behavior\n",
      "consfin_org_count_behavior\n",
      "loans_cash_count\n",
      "latest_one_month_loan\n",
      "latest_three_month_loan\n",
      "latest_six_month_loan\n",
      "history_suc_fee\n",
      "history_fail_fee\n",
      "latest_one_month_suc\n",
      "latest_one_month_fail\n",
      "loans_long_time\n",
      "loans_credit_limit\n",
      "loans_credibility_limit\n",
      "loans_org_count_current\n",
      "loans_product_count\n",
      "loans_max_limit\n",
      "loans_avg_limit\n",
      "consfin_credit_limit\n",
      "consfin_credibility\n",
      "consfin_org_count_current\n",
      "consfin_product_count\n",
      "consfin_max_limit\n",
      "consfin_avg_limit\n",
      "latest_query_day\n",
      "loans_latest_day\n",
      "latest_query_time_month\n",
      "latest_query_time_weekday\n",
      "latest_query_time_day\n",
      "loans_latest_time_month\n",
      "loans_latest_time_weekday\n",
      "loans_latest_time_day\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/wangzhongpu/anaconda/lib/python2.7/site-packages/sklearn/preprocessing/data.py:617: DataConversionWarning: Data with input dtype int64, float64 were all converted to float64 by StandardScaler.\n",
      "  return self.partial_fit(X, y)\n",
      "/Users/wangzhongpu/anaconda/lib/python2.7/site-packages/sklearn/base.py:462: DataConversionWarning: Data with input dtype int64, float64 were all converted to float64 by StandardScaler.\n",
      "  return self.fit(X, **fit_params).transform(X)\n"
     ]
    }
   ],
   "source": [
    "# 接下来对类别特征和数值特征进行填充\n",
    "# 数值特征和类别特征均用用中值进行填充\n",
    "# 缺失值特征特别大的特征‘student_feature’用‘-1’填充\n",
    "for i in data_temp.columns:\n",
    "    temp = data_temp[i].isnull().sum()\n",
    "    if temp:\n",
    "        print i\n",
    "        data_temp[i].fillna(data_temp[i].median(), inplace = True)\n",
    "\n",
    "# data_temp = data_temp.values\n",
    "        \n",
    "# 数值特征归一化 \n",
    "# 从sklearn.preprocessing导入StandardScaler  \n",
    "from sklearn.preprocessing import StandardScaler  \n",
    "# 标准化数据，保证每个维度的特征数据方差为1，均值为0，使得预测结果不会被某些维度过大的特征值而主导  \n",
    "ss = StandardScaler()  \n",
    "# fit_transform()先拟合数据，再标准化  \n",
    "data_temp = ss.fit_transform(data_temp)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0.98850575 1.         0.98850575 ... 0.98850575 0.98850575 1.        ]\n",
      "(4754,)\n",
      "(4754, 1)\n",
      "[1 0 1 ... 1 1 0]\n",
      "(4754,)\n",
      "(4754, 1)\n"
     ]
    }
   ],
   "source": [
    "a5 = data['miss_rate']\n",
    "b5 = a5.as_matrix()\n",
    "print b5\n",
    "print b5.shape\n",
    "b5 = b5.reshape(len(b5), 1)\n",
    "print b5.shape\n",
    "\n",
    "a6 = data['miss_number']\n",
    "b6 = a6.as_matrix()\n",
    "print b6\n",
    "print b6.shape\n",
    "b6 = b6.reshape(len(b6), 1)\n",
    "print b6.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[1. 0. 0.]\n",
      " [0. 1. 0.]\n",
      " [1. 0. 0.]\n",
      " ...\n",
      " [1. 0. 0.]\n",
      " [1. 0. 0.]\n",
      " [0. 1. 0.]]\n",
      "[[0. 0. 1. 0. 0.]\n",
      " [0. 0. 0. 1. 0.]\n",
      " [1. 0. 0. 0. 0.]\n",
      " ...\n",
      " [0. 0. 1. 0. 0.]\n",
      " [0. 0. 0. 1. 0.]\n",
      " [0. 1. 0. 0. 0.]]\n",
      "[[1. 0. 0. 0. 0.]\n",
      " [1. 0. 0. 0. 0.]\n",
      " [1. 0. 0. 0. 0.]\n",
      " ...\n",
      " [1. 0. 0. 0. 0.]\n",
      " [1. 0. 0. 0. 0.]\n",
      " [1. 0. 0. 0. 0.]]\n",
      "[[1. 0.]\n",
      " [1. 0.]\n",
      " [1. 0.]\n",
      " ...\n",
      " [1. 0.]\n",
      " [1. 0.]\n",
      " [1. 0.]]\n",
      "(4754, 3) (4754, 5) (4754, 5) (4754, 2)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/wangzhongpu/anaconda/lib/python2.7/site-packages/sklearn/preprocessing/_encoders.py:363: FutureWarning: The handling of integer data will change in version 0.22. Currently, the categories are determined based on the range [0, max(values)], while in the future they will be determined based on the unique values.\n",
      "If you want the future behaviour and silence this warning, you can specify \"categories='auto'\".\n",
      "In case you used a LabelEncoder before this OneHotEncoder to convert the categories to integers, then you can now use the OneHotEncoder directly.\n",
      "  warnings.warn(msg, FutureWarning)\n"
     ]
    }
   ],
   "source": [
    "# 类别特征one-hot编码\n",
    "\n",
    "a1 = data['student_feature']\n",
    "#print a\n",
    "a1.fillna(-1, inplace = True)\n",
    "# print a\n",
    "b1 = a1.as_matrix()\n",
    "# print b.shape (4754,)\n",
    "\n",
    "from sklearn.preprocessing import LabelEncoder\n",
    "from sklearn.preprocessing import OneHotEncoder\n",
    "from numpy import array\n",
    "from numpy import argmax\n",
    "label_encoder1 = LabelEncoder()\n",
    "integer_encoded1 = label_encoder1.fit_transform(b1)\n",
    "# print(integer_encoded)\n",
    "# binary encode\n",
    "onehot_encoder1 = OneHotEncoder(sparse=False)\n",
    "integer_encoded1 = integer_encoded1.reshape(len(integer_encoded1), 1)\n",
    "onehot_encoded1 = onehot_encoder1.fit_transform(integer_encoded1)\n",
    "print(onehot_encoded1)\n",
    "\n",
    "a2 = data['regional_mobility']\n",
    "#print a\n",
    "a2.fillna(data['regional_mobility'].median(), inplace = True)\n",
    "# print a\n",
    "b2 = a2.as_matrix()\n",
    "# print b.shape (4754,)\n",
    "\n",
    "label_encoder2 = LabelEncoder()\n",
    "integer_encoded2 = label_encoder2.fit_transform(b2)\n",
    "# print(integer_encoded)\n",
    "# binary encode\n",
    "onehot_encoder2 = OneHotEncoder(sparse=False)\n",
    "integer_encoded2 = integer_encoded2.reshape(len(integer_encoded2), 1)\n",
    "onehot_encoded2 = onehot_encoder2.fit_transform(integer_encoded2)\n",
    "print(onehot_encoded2)\n",
    "\n",
    "a3 = data['reg_preference_for_trad']\n",
    "#print a\n",
    "a3.fillna(data['reg_preference_for_trad'].max(), inplace = True)\n",
    "# print a\n",
    "b3 = a3.as_matrix()\n",
    "# print b.shape (4754,)\n",
    "\n",
    "label_encoder3 = LabelEncoder()\n",
    "integer_encoded3 = label_encoder3.fit_transform(b3)\n",
    "# print(integer_encoded)\n",
    "# binary encode\n",
    "onehot_encoder3 = OneHotEncoder(sparse=False)\n",
    "integer_encoded3 = integer_encoded3.reshape(len(integer_encoded3), 1)\n",
    "onehot_encoded3 = onehot_encoder3.fit_transform(integer_encoded3)\n",
    "print(onehot_encoded3)\n",
    "\n",
    "a4 = data['is_high_user']\n",
    "# print a4\n",
    "a4.fillna(data['is_high_user'].max(), inplace = True)\n",
    "# print a\n",
    "b4 = a4.as_matrix()\n",
    "# print b.shape (4754,)\n",
    "\n",
    "label_encoder4 = LabelEncoder()\n",
    "integer_encoded4 = label_encoder4.fit_transform(b4)\n",
    "# print(integer_encoded)\n",
    "# binary encode\n",
    "onehot_encoder4 = OneHotEncoder(sparse=False)\n",
    "integer_encoded4 = integer_encoded4.reshape(len(integer_encoded4), 1)\n",
    "onehot_encoded4 = onehot_encoder4.fit_transform(integer_encoded4)\n",
    "print(onehot_encoded4)\n",
    "\n",
    "print onehot_encoded1.shape, onehot_encoded2.shape, onehot_encoded3.shape, onehot_encoded4.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 特征工程"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 特征选择"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 使用IV值进行特征选择"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(26,)\n",
      "[u'low_volume_percent' u'historical_trans_amount' u'rank_trad_1_month'\n",
      " u'trans_amount_3_month' u'top_trans_count_last_1_month'\n",
      " u'avg_price_last_12_month' u'avg_price_top_last_12_valid_month'\n",
      " u'trans_top_time_last_1_month' u'consume_top_time_last_1_month'\n",
      " u'trans_fail_top_count_enum_last_1_month'\n",
      " u'trans_fail_top_count_enum_last_6_month'\n",
      " u'trans_fail_top_count_enum_last_12_month'\n",
      " u'max_cumulative_consume_later_1_month'\n",
      " u'pawns_auctions_trusts_consume_last_1_month' u'trans_day_last_12_month'\n",
      " u'apply_score' u'loans_score' u'loans_overdue_count'\n",
      " u'latest_six_month_loan' u'history_fail_fee' u'latest_one_month_suc'\n",
      " u'latest_one_month_fail' u'consfin_credit_limit' u'consfin_max_limit'\n",
      " u'consfin_avg_limit' u'latest_query_day']\n"
     ]
    }
   ],
   "source": [
    "import pandas as pd\n",
    "iv = pd.read_csv(\"data/iv.csv\",encoding='gbk')\n",
    "iv_temp = iv[iv.iloc[:,1] > 0.02]\n",
    "# print iv\n",
    "# iv_temp = iv_temp[iv_temp.iloc[:,1] < 0.5]\n",
    "# iv_temp.iloc[:,0]\n",
    "feature_iv =  iv_temp.iloc[:,0].as_matrix()\n",
    "print feature_iv.shape\n",
    "print feature_iv"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 使用RandomForest进行特征选择"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(4754, 99)\n",
      "False\n",
      "(4754,)\n"
     ]
    }
   ],
   "source": [
    "# 特征矩阵X_RandomForest\n",
    "import numpy as np\n",
    "X_RandomForest = np.hstack([data_temp, onehot_encoded1, onehot_encoded2, onehot_encoded3, onehot_encoded4, b5, b6])\n",
    "print X_RandomForest.shape\n",
    "# 预测变量y\n",
    "y_RandomForest = data['status']\n",
    "print pd.isnull(y_RandomForest).any()\n",
    "# y.fillna(data['status'].max(), inplace = True)\n",
    "# 转换为numpy数据的必要性。因为Series数据自带index；而如果index是数字的话，当它切分的时候，会根据index的值划分，如果被划分了一次，\n",
    "# 某些index只会在train或者test中，这样你再次划分train或者test，遇到和之前不在train和test中的index会返回nan。\n",
    "y_RandomForest = y_RandomForest.as_matrix()\n",
    "print y_RandomForest.shape\n",
    "\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "\n",
    "X_RandomForest_DataFrame = pd.DataFrame(X_RandomForest)\n",
    "\n",
    "rf = RandomForestClassifier()\n",
    "rf.fit(X_RandomForest, y_RandomForest)\n",
    " \n",
    "rf_imp = pd.Series(rf.feature_importances_, index = X_RandomForest_DataFrame.columns).sort_values(ascending=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Int64Index([57, 24, 49, 59, 45, 36, 26, 34, 28, 73, 15,  8, 66, 10, 35, 72, 13,\n",
      "            17, 32, 31,  7,  4,  6, 11,  9, 25, 20, 12, 74, 71],\n",
      "           dtype='int64')\n"
     ]
    }
   ],
   "source": [
    "# 筛选 重要性前30个特征\n",
    "feature_rf = rf_imp.index[:30]\n",
    "print feature_rf"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<a id='1.6'></a>\n",
    "## 构造特征矩阵"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 108,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(4754, 82)\n",
      "(4754, 99)\n",
      "False\n",
      "(4754,)\n"
     ]
    }
   ],
   "source": [
    "# 特征矩阵X\n",
    "print data_temp.shape\n",
    "import numpy as np\n",
    "X = np.hstack([data_temp, onehot_encoded1, onehot_encoded2, onehot_encoded3, onehot_encoded4, b5, b6])\n",
    "print X.shape\n",
    "# 预测变量y\n",
    "y = data['status']\n",
    "print pd.isnull(y).any()\n",
    "# y.fillna(data['status'].max(), inplace = True)\n",
    "# 转换为numpy数据的必要性。因为Series数据自带index；而如果index是数字的话，当它切分的时候，会根据index的值划分，如果被划分了一次，\n",
    "# 某些index只会在train或者test中，这样你再次划分train或者test，遇到和之前不在train和test中的index会返回nan。\n",
    "y = y.as_matrix()\n",
    "print y.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(4754, 30)\n",
      "(4754,)\n"
     ]
    }
   ],
   "source": [
    "# 选出有用特征\n",
    "feature_ = list(set(feature_rf))\n",
    "X = X_RandomForest_DataFrame[feature_]\n",
    "X = X.as_matrix()\n",
    "print X.shape\n",
    "y = y_RandomForest.copy()\n",
    "print y.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "<a id='1.7'></a>\n",
    "## 划分训练集和测试集"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "StratifiedShuffleSplit(n_splits=1, random_state=1, test_size=0.3,\n",
      "            train_size=None)\n",
      "('Train Index:', array([4151,  381,  104, ..., 3500,  278,  961]), ',Test Index:', array([2593, 2388, 3542, ..., 3250,  377, 2418]))\n",
      "False False\n"
     ]
    }
   ],
   "source": [
    "# 划分训练集测试集\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "# X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, random_state=23)\n",
    "\n",
    "import numpy as np\n",
    "from sklearn.model_selection import StratifiedShuffleSplit\n",
    "sss = StratifiedShuffleSplit(n_splits=1,test_size=0.3,random_state=1)\n",
    "sss.get_n_splits(X, y)\n",
    "print(sss)\n",
    "\n",
    "for train_index,test_index in sss.split(X,y):\n",
    "    print(\"Train Index:\",train_index,\",Test Index:\",test_index)\n",
    "    X_train, X_test=X[train_index],X[test_index]\n",
    "    y_train, y_test=y[train_index],y[test_index]\n",
    "    \n",
    "    print pd.isnull(y_train).any(), pd.isnull(y_test).any()\n",
    "    # print(X_train,X_test,y_train,y_test)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<a id='2'></a>\n",
    "# 模型选择"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<a id='2.1'></a>\n",
    "## LR模型"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def model_metrics(x_train, y_train, x_test, y_test, y_train_pred, y_test_pred, y_train_proba, y_test_proba):\n",
    "    from sklearn.metrics import accuracy_score, precision_score, recall_score, f1_score, roc_auc_score, auc, roc_curve\n",
    "    %matplotlib inline\n",
    "    import matplotlib.pyplot as plt\n",
    "    \n",
    "    print '[准确率]'\n",
    "    print '训练集：{:.4f}'.format(accuracy_score(y_train, y_train_pred))\n",
    "    print '测试集：{:.4f}'.format(accuracy_score(y_test, y_test_pred))\n",
    "    print '[精准率]'\n",
    "    print '训练集：{:.4f}'.format(precision_score(y_train, y_train_pred))\n",
    "    print '测试集：{:.4f}'.format(precision_score(y_test, y_test_pred))\n",
    "    print '[召回率]'\n",
    "    print '训练集：{:.4f}'.format(recall_score(y_train, y_train_pred))\n",
    "    print '测试集：{:.4f}'.format(recall_score(y_test, y_test_pred))\n",
    "    print '[f1 score]'\n",
    "    print '训练集：{:.4f}'.format(f1_score(y_train, y_train_pred))\n",
    "    print '测试集：{:.4f}'.format(f1_score(y_test, y_test_pred))\n",
    "    print '[auc]'\n",
    "    print '训练集：{:.4f}'.format(roc_auc_score(y_train, y_train_proba)) # auc函数也可以\n",
    "    print '测试集：{:.4f}'.format(roc_auc_score(y_test, y_test_proba))\n",
    "    \n",
    "    print '[roc曲线]'\n",
    "    fpr_train, tpr_train, thresholds_train = roc_curve(y_train, y_train_proba, pos_label = 1)\n",
    "    fpr_test, tpr_test, thresholds_test = roc_curve(y_test, y_test_proba, pos_label = 1)\n",
    "    \n",
    "    label = [\"Train - AUC:{:.4f}\".format(auc(fpr_train, tpr_train)), \n",
    "             \"Test - AUC:{:.4f}\".format(auc(fpr_test, tpr_test))]\n",
    "    \n",
    "    plt.plot(fpr_train,tpr_train)\n",
    "    plt.plot(fpr_test,tpr_test)\n",
    "    plt.plot([0, 1], [0, 1], 'd--')\n",
    "    plt.xlabel('False Positive Rate')\n",
    "    plt.ylabel('True Positive Rate')\n",
    "    plt.legend(label, loc = 4)\n",
    "    plt.title(\"ROC curve\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/wangzhongpu/anaconda/lib/python2.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[准确率]\n",
      "训练集：0.7950\n",
      "测试集：0.7975\n",
      "[精准率]\n",
      "训练集：0.6947\n",
      "测试集：0.6971\n",
      "[召回率]\n",
      "训练集：0.3269\n",
      "测试集：0.3408\n",
      "[f1 score]\n",
      "训练集：0.4446\n",
      "测试集：0.4578\n",
      "[auc]\n",
      "训练集：0.7896\n",
      "测试集：0.7830\n",
      "[roc曲线]\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEWCAYAAACJ0YulAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nFCRJwxcmKGkiB/pq6wAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.linear_model import LogisticRegression\n",
    "\n",
    "lr = LogisticRegression()\n",
    "lr.fit(X_train, y_train)\n",
    "\n",
    "# 预测\n",
    "y_train_pred = lr.predict(X_train)\n",
    "y_test_pred = lr.predict(X_test)\n",
    "# 预测概率\n",
    "y_train_proba = lr.predict_proba(X_train)[:, 1]\n",
    "y_test_proba = lr.predict_proba(X_test)[:, 1]\n",
    "# 模型评估\n",
    "model_metrics(X_train, y_train, X_test, y_test, y_train_pred, y_test_pred, y_train_proba, y_test_proba)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<a id='2.2'></a>\n",
    "## SVM模型"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[准确率]\n",
      "训练集：0.8665\n",
      "测试集：0.7877\n",
      "[精准率]\n",
      "训练集：0.9413\n",
      "测试集：0.7068\n",
      "[召回率]\n",
      "训练集：0.4994\n",
      "测试集：0.2626\n",
      "[f1 score]\n",
      "训练集：0.6526\n",
      "测试集：0.3829\n",
      "[auc]\n",
      "训练集：0.9436\n",
      "测试集：0.7303\n",
      "[roc曲线]\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEWCAYAAACJ0YulAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nUXDOOZfmRcE551za/wEkJVEZXqXRRgAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.svm import SVC\n",
    "\n",
    "svc = SVC(C=1.0, kernel='rbf', gamma=0.1, probability=True)\n",
    "svc.fit(X_train, y_train)\n",
    "\n",
    "# 预测\n",
    "y_train_pred = svc.predict(X_train)\n",
    "y_test_pred = svc.predict(X_test)\n",
    "# 预测概率\n",
    "y_train_proba = svc.predict_proba(X_train)[:, 1]\n",
    "y_test_proba = svc.predict_proba(X_test)[:, 1]\n",
    "# 模型评估\n",
    "model_metrics(X_train, y_train, X_test, y_test, y_train_pred, y_test_pred, y_train_proba, y_test_proba)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[准确率]\n",
      "训练集：0.7857\n",
      "测试集：0.7828\n",
      "[精准率]\n",
      "训练集：0.7773\n",
      "测试集：0.7791\n",
      "[召回率]\n",
      "训练集：0.2048\n",
      "测试集：0.1872\n",
      "[f1 score]\n",
      "训练集：0.3242\n",
      "测试集：0.3018\n",
      "[auc]\n",
      "训练集：0.7899\n",
      "测试集：0.7875\n",
      "[roc曲线]\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEWCAYAAACJ0YulAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nRK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# LinearSVC模型\n",
    "lin_svc = SVC(kernel='linear', probability=True)\n",
    "lin_svc.fit(X_train,y_train)\n",
    "\n",
    "# 预测\n",
    "y_train_pred = lin_svc.predict(X_train)\n",
    "y_test_pred = lin_svc.predict(X_test)\n",
    "# 预测概率\n",
    "y_train_proba = lin_svc.predict_proba(X_train)[:, 1]\n",
    "y_test_proba = lin_svc.predict_proba(X_test)[:, 1]\n",
    "# 模型评估\n",
    "model_metrics(X_train, y_train, X_test, y_test, y_train_pred, y_test_pred, y_train_proba, y_test_proba)\n",
    "\n",
    "# print y_train[0:100]\n",
    "# print y_test_pred[0:100]\n",
    "# print y_test_pred[100:200]\n",
    "\n",
    "# SVM预测结果都是同一个值，可能原因有：1. 可能是由于样本数据没有归一化导致的。由于维度太大，如果不采用归一化处理的话，各个点的距离值将非常大，\n",
    "# 故模型对于待预测点的预测结果值都判为同一个值。2. 也有可能是参数的问题。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<a id='2.3'></a>\n",
    "## DT模型"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[准确率]\n",
      "训练集：0.7018\n",
      "测试集：0.6734\n",
      "[精准率]\n",
      "训练集：0.4427\n",
      "测试集：0.4121\n",
      "[召回率]\n",
      "训练集：0.7269\n",
      "测试集：0.7067\n",
      "[f1 score]\n",
      "训练集：0.5503"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/wangzhongpu/anaconda/lib/python2.7/site-packages/sklearn/tree/tree.py:282: DeprecationWarning: The min_impurity_split parameter is deprecated and will be removed in version 0.21. Use the min_impurity_decrease parameter instead.\n",
      "  DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "测试集：0.5206\n",
      "[auc]\n",
      "训练集：0.7829\n",
      "测试集：0.7358\n",
      "[roc曲线]\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEWCAYAAACJ0YulAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nHhYUjDHGeFhQMMYY4/H/wkPYuvP58iIAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.tree import DecisionTreeClassifier\n",
    "\n",
    "dt = DecisionTreeClassifier(class_weight='balanced', criterion='gini', max_depth=4,\n",
    "            max_features=None, max_leaf_nodes=None,\n",
    "            min_impurity_split=1e-07, min_samples_leaf=1,\n",
    "            min_samples_split=2, min_weight_fraction_leaf=0.0,\n",
    "            presort=False, random_state=None, splitter='best')\n",
    "\n",
    "dt.fit(X_train, y_train)\n",
    "\n",
    "# 预测\n",
    "y_train_pred = dt.predict(X_train)\n",
    "y_test_pred = dt.predict(X_test)\n",
    "# 预测概率\n",
    "y_train_proba = dt.predict_proba(X_train)[:, 1]\n",
    "y_test_proba = dt.predict_proba(X_test)[:, 1]\n",
    "# 模型评估\n",
    "model_metrics(X_train, y_train, X_test, y_test, y_train_pred, y_test_pred, y_train_proba, y_test_proba)\n",
    "\n",
    "# print y_test_pred[0:100]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<a id='2.4'></a>\n",
    "## XGBoost模型"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[准确率]\n",
      "训练集：0.8419\n",
      "测试集：0.7905\n",
      "[精准率]\n",
      "训练集：0.8472\n",
      "测试集：0.6482\n",
      "[召回率]\n",
      "训练集：0.4515\n",
      "测试集：0.3603\n",
      "[f1 score]\n",
      "训练集：0.5891\n",
      "测试集：0.4632\n",
      "[auc]\n",
      "训练集：0.8921\n",
      "测试集：0.7892\n",
      "[roc曲线]\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEWCAYAAACJ0YulAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nMW6WFIwxxrhZUjDGGONmScEYY4zb/wOrOQ3ytgYtkwAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import xgboost as xgb\n",
    "from xgboost.sklearn import XGBClassifier\n",
    "\n",
    "xgb = XGBClassifier()\n",
    "xgb.fit(X_train, y_train)\n",
    "\n",
    "# 预测\n",
    "y_train_pred = xgb.predict(X_train)\n",
    "y_test_pred = xgb.predict(X_test)\n",
    "# 预测概率\n",
    "y_train_proba = xgb.predict_proba(X_train)[:, 1]\n",
    "y_test_proba = xgb.predict_proba(X_test)[:, 1]\n",
    "# 模型评估\n",
    "model_metrics(X_train, y_train, X_test, y_test, y_train_pred, y_test_pred, y_train_proba, y_test_proba)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "[准确率]\n",
    "训练集：0.8479\n",
    "测试集：0.7898\n",
    "[精准率]\n",
    "训练集：0.8553\n",
    "测试集：0.6343\n",
    "[召回率]\n",
    "训练集：0.4743\n",
    "测试集：0.3827\n",
    "[f1 score]\n",
    "训练集：0.6102\n",
    "测试集：0.4774\n",
    "[auc]\n",
    "训练集：0.9089\n",
    "测试集：0.7924\n",
    "[roc曲线]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<a id='2.5'></a>\n",
    "## LightGBM模型"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from lightgbm import LGBMClassifier\n",
    "\n",
    "lgb = LGBMClassifier()\n",
    "lgb.fit(X_train, y_train)\n",
    "\n",
    "# 预测\n",
    "y_train_pred = lgb.predict(X_train)\n",
    "y_test_pred = lgb.predict(X_test)\n",
    "# 预测概率\n",
    "y_train_proba = lgb.predict_proba(X_train)[:, 1]\n",
    "y_test_proba = lgb.predict_proba(X_test)[:, 1]\n",
    "# 模型评估\n",
    "model_metrics(X_train, y_train, X_test, y_test, y_train_pred, y_test_pred, y_train_proba, y_test_proba)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "# 模型评估"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "评估结果\n",
    "\n",
    "| 模型 | 准确率 | 精准率 | 召回率 | F1 score | AUC | ROC |\n",
    "| :------ | :------ | :------ | :------ | :------ | :------ | :------ |\n",
    "| lr | 0.8025; 0.8038 | 0.7051; 0.6990 | 0.3665; 0.3827 | 0.4823; 0.4946 | 0.8105; 0.7897 | ![lr](images/lr.png) |\n",
    "| svm_linear | 0.7920; 0.7926 | 0.7424; 0.7385 | 0.2623; 0.2682 | 0.3876; 0.3934 | 0.8088; 0.7926 | ![svm_linear](images/linear_svm.png) |\n",
    "| svm_rbf | 0.9940; 0.7519 | 1.0000; 0.7000 | 0.9760; 0.0196 | 0.9879; 0.0380 | 1.0000; 0.6997 | ![svm_rbf](images/svm.png) |\n",
    "| dt | 0.7027; 0.6685 | 0.4444; 0.4080 | 0.7377; 0.7123 | 0.5547; 0.5188 | 0.7832; 0.7207 | ![dt](images/dt.png) |\n",
    "| xgb | 0.8479; 0.7898 | 0.8553; 0.6343 | 0.4743; 0.3827 | 0.6102; 0.4774 | 0.9089; 0.7924 | ![xgb](images/xgb.png) |\n",
    "| lgb | 0.8479; 0.7898 | 0.8553; 0.6343 | 0.4743; 0.3827 | 0.6102; 0.4774 | 0.9089; 0.7924 | ![xgb](images/xgb.png) |"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "# 模型调优"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "交叉验证和网格搜索：\n",
    "\n",
    "* 交叉验证：将拿到的数据，分为训练集和验证集（注意这里的数据是在训练集中进行划分的，也就是将原始数据划分得到的训练集再次划分为训练集和验证集），交叉验证一般结合网格搜索使用。\n",
    "* 网格搜素：对模型预设几种超参数组合。每组超参数都采用交叉验证来进行评估。最后选出最优参数组合建立模型。\n",
    "\n",
    "[模型的选择与调优：交叉验证与网格搜索](https://mp.weixin.qq.com/s/i4uELGZFdqhHY8aVRVklDA)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 422,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.model_selection import GridSearchCV\n",
    "\n",
    "# GridSearchCV(estimator, param_grid=None,cv=None)\n",
    "# 对估计器的指定参数值进行详尽搜索\n",
    "\n",
    "# estimator：估计器对象 就是哪一个模型\n",
    "# param_grid：估计器参数(dict){“n_neighbors”:[1,3,5]}\n",
    "# cv：指定几折交叉验证\n",
    "\n",
    "# fit：输入训练数据\n",
    "# score：准确率\n",
    "\n",
    "# 结果分析：\n",
    "# best_score_:在交叉验证中测试的最好结果\n",
    "# best_estimator_：最好的参数模型\n",
    "# cv_results_:每次交叉验证后的测试集准确率结果和训练集准确率结果\n",
    "    \n",
    "# lr\n",
    "# svc\n",
    "# lin_svc\n",
    "# dt\n",
    "# xgb\n",
    "\n",
    "\n",
    "# 参数\n",
    "lr_param = {\n",
    "    'C': [0.01, 0.05, 0.1, 0.5, 1, 5, 10, 100],\n",
    "    'penalty': ['l1', 'l2']\n",
    "}\n",
    "svc_param = {\n",
    "    'gamma': [0.01, 0.05, 0.1, 0.5, 1, 5, 10], \n",
    "    'C': [0.01, 0.05, 0.1, 0.5, 1],\n",
    "    'class_weight': ['balanced',None] \n",
    "}\n",
    "lin_svc_param = {\n",
    "    'C': [0.01, 0.05, 0.1, 0.5, 1]\n",
    "}\n",
    "dt_param = {\n",
    "    'max_depth': [m for m in range(3,8)], \n",
    "    'min_sample_split': range(2, 100, 5), \n",
    "    'class_weight': ['balanced',None]\n",
    "}\n",
    "xgb_param = {\n",
    "    \"max_depth\": [10,30,50],\n",
    "    \"min_child_weight\": [1,3,6],\n",
    "    \"n_estimators\": [200],\n",
    "    \"learning_rate\": [0.05, 0.1,0.16]\n",
    "}\n",
    "lgb_param = {\n",
    "    \"max_depth\": [5,10, 15],\n",
    "    \"learning_rate\" : [0.01,0.05,0.1],\n",
    "    \"num_leaves\": [30,90,120],\n",
    "    \"n_estimators\": [20]\n",
    "}\n",
    "scoring = 'roc_auc'  # 评价标准\n",
    "n_fold = 5  # n折交叉验证"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## LR模型"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 160,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "最优分数： 0.788989992938\n",
      "最优参数： {'penalty': 'l1', 'C': 0.05}\n"
     ]
    }
   ],
   "source": [
    "lr_grid = GridSearchCV(lr, lr_param, cv=n_fold, scoring=scoring)\n",
    "lr_grid.fit(X_train, y_train)\n",
    "print '最优分数：', lr_grid.best_score_ # 最优分数\n",
    "print '最优参数：', lr_grid.best_params_ # 最优参数\n",
    "# print '交叉验证结果：', lr_grid.cv_results_ # 结果\n",
    " \n",
    "# 把参数重新赋值给原模型\n",
    "# lr.set_params(**lr_grid.best_params_)\n",
    "# lr.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 483,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[准确率]\n",
      "训练集：0.7971\n",
      "测试集：0.8024\n",
      "[精准率]\n",
      "训练集：0.7198\n",
      "测试集：0.7262\n",
      "[召回率]\n",
      "训练集：0.3138\n",
      "测试集：0.3408\n",
      "[f1 score]\n",
      "训练集：0.4370\n",
      "测试集：0.4639\n",
      "[auc]\n",
      "训练集：0.8007\n",
      "测试集：0.7942\n",
      "[roc曲线]\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYIAAAEWCAYAAABrDZDcAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "lr_ = LogisticRegression(C = 0.05, penalty = 'l1')\n",
    "lr_.fit(X_train, y_train)\n",
    "# 预测\n",
    "y_train_pred = lr_.predict(X_train)\n",
    "y_test_pred = lr_.predict(X_test)\n",
    "# 预测概率\n",
    "y_train_proba = lr_.predict_proba(X_train)[:, 1]\n",
    "y_test_proba = lr_.predict_proba(X_test)[:, 1]\n",
    "# 模型评估\n",
    "model_metrics(X_train, y_train, X_test, y_test, y_train_pred, y_test_pred, y_train_proba, y_test_proba)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## SVM模型"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 312,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "最优分数： 0.774302939611\n",
      "最优参数： {'C': 0.01, 'gamma': 0.01, 'class_weight': None}\n"
     ]
    }
   ],
   "source": [
    "svc_grid = GridSearchCV(svc, svc_param, cv=n_fold, scoring=scoring)\n",
    "svc_grid.fit(X_train, y_train)\n",
    "print '最优分数：', svc_grid.best_score_ # 最优分数\n",
    "print '最优参数：', svc_grid.best_params_ # 最优参数\n",
    "# print '交叉验证结果：', svc_grid.cv_results_ # 结果"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 313,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "最优分数： 0.781479503689\n",
      "最优参数： {'C': 0.01}\n"
     ]
    }
   ],
   "source": [
    "lin_svc_grid = GridSearchCV(lin_svc, lin_svc_param, cv=n_fold, scoring=scoring)\n",
    "lin_svc_grid.fit(X_train, y_train)\n",
    "print '最优分数：', lin_svc_grid.best_score_ # 最优分数\n",
    "print '最优参数：', lin_svc_grid.best_params_ # 最优参数\n",
    "# print '交叉验证结果：', lin_svc_grid.cv_results_ # 结果"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 324,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[准确率]\n",
      "训练集：0.7490\n",
      "测试集：0.7491\n",
      "[精准率]\n",
      "训练集：0.0000\n",
      "测试集：0.0000\n",
      "[召回率]\n",
      "训练集：0.0000\n",
      "测试集：0.0000\n",
      "[f1 score]\n",
      "训练集：0.0000\n",
      "测试集：0.0000\n",
      "[auc]\n",
      "训练集：0.8602\n",
      "测试集：0.7820\n",
      "[roc曲线]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/wangzhongpu/anaconda/lib/python2.7/site-packages/sklearn/metrics/classification.py:1113: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples.\n",
      "  else:\n",
      "/Users/wangzhongpu/anaconda/lib/python2.7/site-packages/sklearn/metrics/classification.py:1113: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 due to no predicted samples.\n",
      "  else:\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYIAAAEWCAYAAABrDZDcAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "svc_ = SVC(C=0.01, kernel='rbf', gamma=0.01, class_weight=None, probability=True)\n",
    "svc_.fit(X_train, y_train)\n",
    "\n",
    "# 预测\n",
    "y_train_pred = svc_.predict(X_train)\n",
    "y_test_pred = svc_.predict(X_test)\n",
    "# 预测概率\n",
    "y_train_proba = svc_.predict_proba(X_train)[:, 1]\n",
    "y_test_proba = svc_.predict_proba(X_test)[:, 1]\n",
    "# 模型评估\n",
    "model_metrics(X_train, y_train, X_test, y_test, y_train_pred, y_test_pred, y_train_proba, y_test_proba)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 336,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[准确率]\n",
      "训练集：0.7896\n",
      "测试集：0.7891\n",
      "[精准率]\n",
      "训练集：0.7974\n",
      "测试集：0.7938\n",
      "[召回率]\n",
      "训练集：0.2168\n",
      "测试集：0.2151\n",
      "[f1 score]\n",
      "训练集：0.3409\n",
      "测试集：0.3385\n",
      "[auc]\n",
      "训练集：0.8071\n",
      "测试集：0.7934\n",
      "[roc曲线]\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYIAAAEWCAYAAABrDZDcAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4yLCBo\ndHRwOi8vbWF0cGxvdGxpYi5vcmcvhp/UCwAAIABJREFUeJzs3XeYVNX5wPHvtJ2yO9sbLG1ph16k\nCiiCWMCKDUGplvyMxiSaGDWWRBM1Ro2aqImFoogiRkUQREUQCYgUEWmHXnZhe2+zU+7vj1mGXXZZ\nBtzZmd05n+fhce695977Xlbm3Xvuue/RaZqGoiiKEr70wQ5AURRFCS6VCBRFUcKcSgSKoihhTiUC\nRVGUMKcSgaIoSphTiUBRFCXMqUSgKIoS5ozBDkBRmpoQ4hCQAriBMuBz4B4pZVmtNiOAvwBDAA+w\nBviDlHJnrTbRwBPAdUA8kAUsBf4ipcxrjmtRlOag7giU1uoqKWUUMAAYCDx0YoMQ4nzgC2Ax0BZI\nB34E/ieE6FzTJgJYCfQGLgeigRFAPjA0UEELIdQvZ0qzU//TKa2alDJLCLECb0I44VngbSnlS7XW\nPSKEGAT8CZhW86cDMKbWnUQO8OTpziWE6A28CAwCnMBLUsqnhBBzgQwp5SM17S4C5ksp29UsHwJe\nA27xLoongQFSyhtqHfslQCelvFcIEQO8AEzAezczB3hcSuk+u78dRfFSdwRKqyaEaAeMB/bVLNvw\n/ma/qIHmHwCX1HweB3xeuzvpDOexA1/h7YZqC3TFe0fhr8nAFUAs8A4woaZrCiGEAbgJWFDTdh7g\nqjnHQOBS4PazOJei1KHuCJTW6hMhhAZEAV8Dj9esj8f7C9DxBvY5DiTWfE4ANp/F+a4EsqSUz9cs\nVwEbzmL/l6WUR2s+HxZCbAGuBd4GxgIVUsrvhBApeBNbrJSyEigXQvwDuBP4z1mcT1F81B2B0lpd\nK6W0AxcBPTj5BV+ItzulTQP7tAFOPATOP02b02kP7D+nSL2OnrK8AO9dAsAUTt4NdARMwHEhRJEQ\noghvAkj+GedWwpxKBEqrJqX8BpgLPFezXA6sB25soPlNnOzO+Qq4TAgR6eepjgJdTrOtHLDVWk5t\noM2pZYAXARfVdG1N5GQiOAo4gEQpZWzNn2gpZW8/41SUelTXkBIOXgQOCSEGSCm3Ag8CK4QQu/E+\naDUC9wPn4x1OCt5++l8A/xVC/AbYA8TVrNsqpVx2yjmWAi/UtH0NiAB6SSk3AFuB+4UQf6lZ/5sz\nBSylzBVCrK6J76CUclfN+uNCiC+A54UQj+IdHpsOtKtJeopy1tQdgdLqSSlz8fa1P1qzvBa4DO/7\nAceBw3gfuo6SUu6taePA+8B4N/AlUAJ8j7eLqV7fv5SyFO+D5qvwvm+wFxhTs/kdvMNTD+EdtrrQ\nz9AX1MSw4JT10/AmlJ14u7o+5Oy6sRSlDp2amEZRFCW8qTsCRVGUMKcSgaIoSphTiUBRFCXMqUSg\nKIoS5lrc8NHc3NJzfrodF2ejsLCiKcMJeeqaw4O65vDwc645KcmuO922sLojMBoNwQ6h2alrDg/q\nmsNDoK45rBKBoiiKUp9KBIqiKGFOJQJFUZQw1+IeFiuKooSjsq0/YIi1QSfR5MdWdwSKoighzuOs\nJvf9BRx8czYeZ3WTHz9gdwRCiNl4J+vIkVL2aWC7DngJ73R7FcAMKeWWQMWjKIrSUhUs+wxnXi5O\noHD5MhKuvrZJjx/IO4K5eCf9Pp3xQLeaP3fiLd2rKIqi1FKdm0Ph5yernhcs/wxnbm6TniNgdwRS\nyjVCiE6NNLkG7wTiGvCdECJWCNFGStnQFIKKoigtTk5RJdsP5GPQ63BqDjyaB6dWza6qjXg0d4P7\nFJQ4MBpO/o4+Zu1u2judvmXN6STnvfmk3fvbJoszmA+L06g7PV9GzbpGE0FcnO1nvVSRlGQ/531b\nKnXN4UFdc9PRNI2MnDIqHS5+3Jtb54s5pzqDY46DAHg8Ho7llWM11/8qrXa6yS2qBMAQm4veVubf\nyS11F5268npNIszGJr32YCaChl53PmP5iJ/zSnlSkp3c3NJz3r8lUtccHtQ1N87l9lB76pXj+eVs\n3ZuHp2bl3oxi3G4PBoOeXYcLAdBZytBH5wOgjyrGmHgMTdOh053yNRUFDUZhApOt7iqb3k6iKRUN\njZ62QUQa6n+ZezwabRIisTrKIToa0grw/OMf4HJ54zKZiL3u5rP+eTeWOIKZCDLwTvh9QjvgWJBi\nURSlBXG5Pcz7bCclpVUA5BRWciyvnEjrya80t74KpzmP7MJK3zpT2j4wNNAlE3vyo7nms95cWa+Z\nTqeRHNEWmyGSfvahNSuhbbwNg7H+I1cdOswR3h4Mm9FKamTyGa9Nc7sp/OoL8j/5iJTpM4kePoK8\n8VdQsGQxAPHjr8CUlHTG45yNYCaCT4F7hBDvA8OAYvV8QFGU2hzVborKHHz4zX4sEQbWb88mOc5K\nVsHJngGdpRwM3j70woTDGMze5KCL8v5mb06sf1yDy4ZOp8Nk1GM0eDsn9Doden3tjgozBr2Ba7qM\nB8BqsCDiu6LXBW6MjePoUbLmzcZx6CAGezR6s7efKH7CFZSuX4fBqCdu/IQmP28gh4++B1wEJAoh\nMoDHAROAlPLfwDK8Q0f34R0+OjNQsSiK0nJ4NI1dhwv5eM0BDmQVoDM6MbbdDy4Xhs6QD9iS9Ljc\nHqLiy6lsoGNGr9OjaTo0NG7odnWdbd3jupAWFVpTPHucTgqWLaVg2VJwu4k+fyRJkyZjiIoCQG+K\nIOnmKcTE2nCbIpr8/IEcNTT5DNs14O5AnV9RlNCmaRrVTg+Hs0vZeaiAaqeHz78/cmIr6N1YBqxG\nZ3TV3xcwACc6b1IjU+ib0BOAHvHd6BHfrTkuocmUbfyegiWLMcbHkzJ1BpF9+9VrEzVgIPEBehak\nSkwoitKkNE2jqtrbD19Q6iCvqJIdhwow1fShu90aX2w8etr99VGFmHtsBL3Ht25wygCGpp5He3ua\nb11CQhT5+WUYdUZsJmuAriZwPA4H6HToIyKwDz8fV2kJsaMvQm9p/mtRiUBRlJ/F5fawbnsWq7Zk\nUlpZTUGJo34jvRt9TC46nffL3RDvXR1pNWHU6+jUNpo8wx7KPAVUek72/4u4rlzWcSwivmu9Q8Za\n7DgjTjvXSkir2LWT7HlziBo0iKQbb0an1xN/2figxaMSgaIoPi63h5yaUTb7jxXj9jQ8oruo1MGR\n7DJKK6vZn1nSYJsBXRPRNA2Hy83hlPcbbOOs+bNLA2p6gGLNMURH2Lmtz60kWuN/5hWFFndFObkf\nLKRk7RrQ69EZjGiahk4X3ISmEoGihJGCkioKyxys2HCETTKXSEvdr4Dyqvr98f4ae14alw7tQHKs\nt2vD6XZS6Cjmo31LIM/b5or0S7BHnH48e4otke5x9X/7bw3KfthC9vy3cRcXYW7fnpTpt2Hp1CnY\nYQEqEShKq+fxaDy/cCvH88spKqtbubK8ykVaUqRvOcYeQVFZJYN7JON2e7DbImib6H0r6mjVPnKc\nmbX21hEdacKg12My6YEiVmbvgGzv1rWZ39U518SuVzCuw+hAXGLIcxw7xrFXXkZnNJJw7XXEXz4B\nnTF0vn5DJxJFUZpUblElbyzZyb7M4nrbLhvanjYJkYzq2wa9Xkelq5J1xzby0b6lAGwE72BvD5DT\nyEnqv3NVz4g2Q4i3xIddEtA0Dc1Rhd5ixdy2LUk334KtVy/MbdPOvHMzU4lAUVqwjNwyDh4rITra\nSmlpFRt357A/09u373B6R+7ozOXo7YWMPa8dHVOj0KEDioAiNmR5f8Ofv3tRneP2iq8/+YmGRpvI\nFEa2Hep3fInWBIz68Puacebnk/3OPHC7SLvv9+h0OuLGXRLssE4r/H5CitLC5RRV8unag6zbntVw\nA2M1elsJlmg9GqDv8j0Aa4u2s7ao8WNP6zmJfkm9sBpb3nDMUKB5PBR/s4rcDxehOaqw9eqNp6oK\ngzW0/z5VIlCUEFRSXs2Xm46SXVhJSXk1+2u6dxoaxRNnNzNtQi9KS6vId2bxRfF7vm21x6JM7XlT\no+cUcV2Js8Q22kY5veqs42TPm0Pl3j3obTaSZ95G9IhRQR8R5A+VCBQlBGiaRqXDzYHjxbyw8MfT\ntkuOs6LX6bCajYw9L42UdlWsP76RPeSQ6cpmb/EBX9urOp+cF6pzTIdWOxonFHic1Rx99mncJSVE\nnTeI5FumYoxpOUlVJQJFCbLl3x1m0er9DW67e2IfEmIspCVGYtDr6xRFK3aU8PD//lFvH3tEFI8N\n+x22U2sgK03O43SiN5m8tYAmTUZnNGIfNCTYYZ01lQgUJUgKSx28t3Ivm3afHJYzoGsiZVVOHpg8\n0DcZiqZprDi8iiUHPsekN3Kiw8fpOTlr1ctXPEFhQTkGnUF17zQDj7OagqVLKN34PR0f+zN6i4Xo\nYecHO6xzphKBojQjh9PN2m3HWbr+EMWnjOl/5bcXYjUbcXlc7C3ch9PjZNnBLzladnKaDqfHRUf7\nyWk8DHoDE7teQWpUEobKU6a2UgKict9esufOpjrrOMaEBJz5eZjT2gU7rJ9FJQJFaQZllU4+/vYA\nq7Zk1ts2Y3wPLujXBp1OR6Wrir9seJ4iR92x/5EmG+cl9+fGbldj0J/7VK3KufNUVZH30YcUrVoJ\nQOzYcSRedwN6S8tPwCoRKEqAuNze+Wz/NGdjrbUamBzceFFX+nZJwG41AVBS7S0t/Oymf/qSwIVp\nI0i0xtM9rkudqptKcBz/z6uU/7QNU2oqqdNvw9qtZZW6boxKBIrSxPZlFvP15gy+25ldZ310ZATJ\n520j03GIpcWrWbrl9MeY2Wsyg1MHBjhS5Uw0txudwXsHFn/VNZjbdyD+qqvRB2BymGBSiUBRmkBh\nZQkr9mxgX2Yhh7PLMLWXWIdqaFrNo10duNCR6fC+BzAouf9pj9UvsZdKAiGgdPMm8hYtJO039xOR\nmoq1cxesnbsEO6yAUIlAUc7RjkMFLFt/mMLqPIrbf+ldaYGIjifbdI1Nr7dfv6ReYVd3pyVxFReR\ns2A+ZZs3oTMaqTpyiIjU1GCHFVAqESjKOah0VfGPpaswJB/B2P64b31HxwX0aJdEh5Qo2tvbktDK\n6um3ZpqmUbLuf+QufA9PRTnWbt1JmT6TiNTQmt84EFQiUJSz8NyHG8iI+A5nVCbmXifXR0fY+b9+\nM+gY3f70OyshrXD5Z+R99CE6s4XkW6YSM3oMOr0+2GE1C5UIFOUMNE1j6/4c/vnxVqyDVtbZ1sMy\nhLGiL70TegQpOuXnqD07WPTIC3AcPULiDTdhSkgMcmTNSyUCRWlEdkkRT2x6CgDroJPrHxj8K9rb\n09DrwuM3xtao+vgxsubNIf7yCUQNGIgxJoY2v/hlsMMKCpUIFKWGx6NRWe3iVy+uIaLH9+itZehM\nJ8s4uIsT6NEhjsvSL1JdQC2Y5nJRsGI5BUsWo7lcVOzcTtSA8B6lpRKBEpbWZn7Hmsz1gDcB5BZX\nUu30AGDu70RvrgLA6IzGo3My3HwtU64d0CJKCiunV3XkMNlz3sJx9AiGmBiSp0zFPmhwsMMKOpUI\nlLCxv+gQXxxehQcPO/MlAAYtApfbAzrQ1bwjpNfpMBusXNXlUi5qNzKIEStNqWL3LjJe+Dt4PESP\nuoCkG2/GEBl55h3DgEoESqvn0Tw43NX879gGtufvOrm+PJrKHSN8y7+6vi8DuyUFI0QlgE48ELZ2\n7UZk/wHEXjSWyN59gh1WSFGJQGnVNE3jvm8ewelx+dZV/TQSzWEDj/dB7yCRxN0T+wYrRCVAPFWV\n5P53EcbYOBKuuAqd0Uja3fcGO6yQpBKB0mrtyN/NikOrTiaB4mRcVRa0yihAR0xkBLdf1YvendRL\nX61N+U/byH5nLq6CAm99oMsn+GoGKfWpRKC0OuWVTp5avIxM+2rfOufxTriOnhzr/8i0wXRuGx2E\n6JRAcpeVkbNwAaXr14HBQPyVVxN/xVUqCZyBSgRKqyKPFPK3BT8QIX7CAGguE4ajA4l1tuFPvxmK\nzWIKdohKgLhKSjj8+CO4S0swd+xE6ozbMLdXw3z9oRKB0ipk5pXz6JsbvAumKgwx+QD88fx7SLu0\n9deKUcAYHU3kgAFEpKQSd8ll6i7gLAQ0EQghLgdeAgzAm1LKZ07Z3gGYB8TWtHlQSrkskDEprcvX\nWzI4nl/Bys0Z6CIqiOi2FX1kiW97m6iUIEanBJKmaRSvXUPlvn2kzpgFQOr0WUGOqmUKWCIQQhiA\nV4BLgAxgoxDiUynlzlrNHgE+kFK+JoToBSwDOgUqJqV1eXHRj2zbnw9oWAZ/iU7v8W2LNcfwqwF3\nqBIQrZQzN5cd/3yB4h+3obdYcF51ddjVB2pKgbwjGArsk1IeABBCvA9cA9ROBBpw4oldDHAMRWlE\neZWTT9ceorSy2psE9C6sg7/ybe9gb8e9I2ZgdaoHwa2R5vFQ9PVX5H30IVp1NZF9+5E8dTqm+IRg\nh9aiBTIRpAFHay1nAMNOafMn4AshxK+ASGDcmQ4aF2fDaDz3vr+kJPs579tStYZrdns0/jpnAxtr\nTf+oi6jE3Gedb/mBUXcxOK1fMMILCa3h59wYTdPY8difKd72E0a7nc73/JLEC0eFXdmPQPycA5kI\nGvrpaKcsTwbmSimfF0KcD7wjhOgjpfQ0sC8AhYUV5xxQUpKd3NzSc96/JWrJ15xXXEluYSXrd2az\ndtvxOtsuHhnNOufnvuW7+99Gx4h0cnNLW/Q1n6twuWaT6IXdYiNp8i0kdU4Li2uu7ef8nBtLIIFM\nBBlA7bFb7ajf9XMbcDmAlHK9EMICJAI5AYxLCXE5RZU88sYGbw2gU1x7QTpXj0zn7q8f8K2b3utm\neiWI5gxRaSZVhw5S8PlyUm+7A73JRNyll4fdHUBzCGQi2Ah0E0KkA5nAzcCUU9ocAS4G5gohegIW\nIDeAMSkhrtrp5sF/r/ct63U6rhrZidioCEb2S+X7rC0slFt92/826nGiIlThsNbGU11N/uKPKfzi\nc9A0KoYNI2rgIJUEAiRgiUBK6RJC3AOswDs0dLaUcocQ4glgk5TyU+B+4A0hxG/xdhvNkFKe2n2k\nhAGX28PWvXm8+sl237rnfjmC+GiLd7vHxZIDn/PVkW982zvHdFRJoBWqkLvJnjcHZ042pqQkUqbN\nxNaz15l3VM5ZQN8jqHknYNkp6x6r9XknoOr8hrmFX+9lxfdH66x76NbziI+2sDn7R746spojpZm+\nbT3iunFJx4voFtu5uUNVAix/yWLyF38MOh1xl1xGwrXXoTebgx1Wq6feLFaCxu3xcMezq0+u0HkY\n3t/OyH5tOO7ZxRc/7vTNG3DCpR3HcFnHMViMluYNVmkWls5diEhrR8r0mVg7dwl2OGFDJQKlWbk9\nHvKKqljx/RFWbz05dmD0gLZsjJjDj2j8uLvuPgmWOB4f/gAGvSoZ0Nq4S0vJ/WgRCVddiyk+nsje\nfbA9/gQ6vXoRsDmpRKA0mczcMvZmFLPnaBE6nY6Gnuut257l+6yzFWOIzeX83qnY7BItw/t4aESb\noQDEW+IY034kJr1JJYFWRtM0SjduIHfBu7jLSjHYbCTdeDOASgJBoBKB0iQOHi/hyXmb/G7fo0Ms\nFe23kes8xqbifVDsXT88dTC39LwhQFEqocBZWEjOu29TvvUHdBERJN44ibhLLgt2WGHNr0QghIgC\nOksptwU4HqWF0DSNT749yA97c7HbIth1uNC37Y4rexEdGUFKvLXBfU1mF29sn0duyTGMOgN39fcW\nCtPrdHSK7tAs8SvBUf7TNo6//hqeykqsogcp02YSkaIKAwbbGRNBTQXRNwA30EkIMQR4VEp5daCD\nU0JTvYe8lPs+1R7yeaoqVxWv//Q2snCfb13/pD70iO8WoEiVUBPRpg06s5nkGycRc8Fo9V5AiPDn\njuAJvDWCPgOQUm4UQqjH+WGo0uFi/Y4s5n+xx7duUPckfnFNb3Q6MJyhb/flH97gcOnJYaJ/GHIv\nHeztAhavEnyax0PhlyuwpHfG1l1gSkwi/em/ozepCYJCiV9dQ1LKY0LUeYW/OjDhKKFo3fbjvLl0\nV731d0/syyCRVG+9pmmsyVxPsaOkzvoih/dBwGPDfkeyLUn9NtjKOTIzyJrzFo5DB7F06UqHhx4B\nUEkgBPmTCMqFEEnUFIwTQlyA79Ge0pplFVTw8Ovf1VlniTAwoGsis67oidFw8g7gSGkGf9v4Mgad\nAbfmPu0xY80xpEQmByxmJfg0l4v8z5ZQsGwpuN3Yh51P8s2nVpdRQok/ieBhvGUi0oUQXwG98M4r\noLRSHo/G3OW7WfvTyYqfI/qkctPYrkTbIhrc563t7wLg1tx0iu6Ay+NiUHJ/usSm12mXZFN141uz\n6twcjv3zJaqPZWKMiyd56jSi+g0IdljKGZwxEdRUBb0YGIW3tPT/pJT5AY9MCYr6D4IbfwBcWl3G\nu7sXUVRVBMAzox7DHhEV6DCVEGWMjkFzu4gZPYbEG27CYG145JgSWvwZNfS8lPJ+YEkD65RWZMGX\ne/hqc4Zv+bYrenJ+n1T0p+nLX7x/OauOfovT4wJgaOp5KgmEoYrdu3AVFRI9fAR6s5mOj/4ZvUWV\nAGlJ/OkaGtPAurFNHYgSHGWVTu596dt66++9vh8Dup1+DtjXfpzD9nzvA2QdOn7Rbzp9E1WFyHDi\nrqgg78MPKF6zGr3VSmT/gRisVpUEWqDTJgIhxPXADUBHIcSCWptigMpAB6YE3hcbj/L+yr2+5UiL\nkb5dErjzqt6N7lfkKPYlgeu7XsnwNoOxmWwBjVUJLWVbfyB7/jzcRUVEpLUjdeZtqhuoBWvsjuAA\nsBIYUfPfE0qALwMZlBJ4+zOL6ySB308eSM+OcX7t+/q2t32fx3a4sMljU0KXx1lN9pzZlH7/HTqj\nkYRrryP+8gnojKpaTUt22p+elPIH4AchxGIppZo1rBXYfjCfect34/ZoFJV5XwWJspp4/u4RmIz+\nFXX7YM8nvpfCHhmmHhOFG53RhKeqEkvnLqTMmIW5bVqwQ1KagD9pvEAIMQsYgHcqSQCklHcGLCql\nyc1etqveBPAAf7lj2GmTgEfzUFBVBGgcKc1k5ZE1HCo5AkDX2HRSbep9gHDgLCigfOsWYseOQ6fT\nkXr7L9BbLKpKaCviTyL4N2AFRgOvA5OBbxrdQwkp+cVVdZLA83ePJM7e8KxPmqax7tj3ZJQdY03m\n+gbbDEruz6w+twQkViV0aB4Pxd9+Q96ihXiqqjB36oy1c2cMNvU8qLXxJxEMB/oBP0opnxRC/AtY\ncIZ9lCBzuT2s357FwlX7qKhy+dbPfrDhAV+bsreSV1nAkgOf19s2NPU89Do9NqOVCenjsBrVQ8HW\nrjo7i+x5c6jcI9FbraRMn4klPf3MOyotkj+JoFJKqQkh3EIIq5SyUAihOgZD2KbdOXUmgT/huV+O\naLD9roI9zNlRN7eP7zSO85L7EWuOwWZSX/zhpPDLFeR99CGa00nkgIGk3DoNY6x/AwmUlsnfZwQx\nwBfAUiFEHqAeHoeo7IKKOkngxou7MbJ3Sr3SEGXOco6WZrL66Fq253vnhow1x3BrjxtpE5VCrDmm\nWeNWQoe7vAy9xUrybXcQNWiIKg4YBvxJBFcDTrw1h6YCscDcAMaknKPyKicP1SoS9+YfxpCSHE1u\nbmmddoVVRTyy7qk66xIscfyy/22kqoJwYcfjdFKydg0xo8eg0+uJv+Jq4sZdhiFKvSUeLvypNXSi\n5LSbmgQghBgPLA9cWMq5+NWLJ98QfuW3FzZYGqLKVcW2vJ2+5as6X0bnmI50j+vaLDEqoaVy/z6y\n586m+vgx0DRix47zlolWpaLDSqOJQAgxEegALJNS7hVCjAP+CsShEkFIKak4OUXEU3cOx2qu+6P9\nPmsL7+3+L9Uep2/dtJ6TGNZmULPFqIQOj8NB3scfUrTyK9A0YsZcTPSIkcEOSwmSxkpM/ANvuekt\nwC+EEB8DdwF/Al5rlugUv2iaxm9eXutbTo231dn23KZXOFhy2LeuR1w37BF2eiZ0b9Y4ldBQsXsX\nWXPfwpWXhyklhZTps7B1F2feUWm1GrsjGA/0l1KWCiFSgUPAACnl7maJTPHb1r15vs+PTh9cZ9t/\ndy73JYERbYZyeaexJFjjmzU+JbS4y8pwFRQQN/4KEq66Bn1Ew3NMKOGjsURQIaUsBZBSZgkh9qgk\nEHqenr+ZvRneCeM6JEeR3iaawqoiSp1loMEH273Vwwcl9+eWnjcEM1QliMp+2IK1W3cMUVFEDRpM\npyefJiIlJdhhKSGisUSQKISoXUYipvaylPL1wIWlnMlmmcsrH//kW05LiuSxGUPYnreL17bNqdd+\nRu/JzRmeEiJcxcXkvDefsk0biR4xitRZt6PT6VQSUOpoLBF8A1xQa3lNrWUNb7kJJQjKKp11ksCg\n7kncfV1fXB5XnSQwtv0F2Gxm+sX0Ra9TdWHCiaZplK5fR877C/BUlGPp0pX48ROCHZYSohqrPjq1\nOQNRzszpcvOPD35k95Ei37o3HxiDXq9jf9EhXtjyqm/9P0b/lQiDiaQke733CJTWzVmQT/bbc6nY\n/hM6s5mkybcQO+ZiVSROOa2AFhEXQlwOvAQYgDellM800OYmvCORNLz1jKYEMqaWauPuHF6r9cZw\nhFHPI9MGo9fr8GieOkngzr7TiTCoceDhylNVReXuXdh69SZl2gxMiUnBDkkJcQFLBEIIA/AKcAmQ\nAWwUQnwqpdxZq0034CFgZE0NI/Va62nUTgK/uq4vA7uf/Mdd5XL4Pp+4E1DCS2XmMRy5JZjT0jC3\nTaPDHx8nol07VR5C8Usg7wiGAvuklAcAhBDv430vYWetNncAr0gpCwGklDkBjKfFenHRj77Pb/1h\nTL1/3G7NDUC/xN4qCYQZze2m8IvP2fvpJ0SktqHDI4+jMxgwt28f7NCUFsTvRCCEiDvxhe2nNOBo\nreUMYNgpbbrXHPt/eLuP/iSlrF8HuZa4OBtGP2fTakhSkv2c9w2Gw1klbNufD8Cw3qkkJ0fXa/P0\n8n8AYLGYGry+lnbNTSEcrrlua1WwAAAgAElEQVTswEH2/etVyvcfwBQbS6cpN5GYGhvssJpVOPyc\nTxWIaz5jIhBCDAYW1bRtX7N8m5TyrjPs2tA9qdbA+bsBFwHtgG+FEH2klEWn7nhCYWHFmUI+rZb4\n4PSev6/yff7FVb3qxO90O/nqyBoySryTzoxKHlHv+lriNf9crf2aPc5qCpYuoeDzZeB2Ez1iFD1+\neTtFVbTq6z5Va/85N+TnXHNjCcSfO4IX8VYgfRtASrlJCDHPj/0ygNr3p+2AYw20+U5K6QQOCiEk\n3sSw0Y/jt2oHj5cwe9ku3/Lf76o7l0BhVRH/3beUH3K2AdAxuj1dYjs1Z4hKkGhOF8X/+xZjbCwp\n02YS2bsPJrsdqsLrS1FpOv4kArOU8ich6tQiqT5d41o2At2EEOlAJnAzcOqIoE/wTn05VwiRiLer\n6IAfx27Vnnl3C3uOnrwpSo61khBjQdM05u9aREbZMTLKTubUYamDmNzj+mCEqjQTT1UVjmOZWDt3\nwWCz0e7X92FKSkZvsZx5Z0U5A38SQbUQwkZNt44Qogd+JAIppUsIcQ+wAm///2wp5Q4hxBPAJinl\npzXbLhVC7MRb5vr3Usr8c7yWVuH/nltNtcvjW37ltxdiifA+E3F6nHyXtQm9To/FYKba42R6z0n0\nT+qDSR/QkcBKEJXv2E7223PwVFXR6YmnMMbEYG7fIdhhKa2IP98eTwFfAm2FEG8CVwAz/Dm4lHIZ\nsOyUdY/V+qwB99X8CWseTeP2v518HnD50A7cNPbkHAFFjmL+/aP3reFusZ25d+Cd9Y6htC7usjJy\nP3ifknVrQa8n/vIJ6G1q2lCl6fkzMc1nQog9wOV4HwA/K6XcE/DIwkhxmYPf/ut/vuUrR3Tkugu7\n1Gnz6f7POVrTHdQpWv022NqVbt5Izrvv4C4pwdyhIykzZmHp0DHYYSmtlD+jhiYDH0kp/9kM8YSl\nt2o9FL73hn4M6JroW86vLOTbzPVsyNoMwB19pjIguW+zx6g0H03TKPrqSzwVFSRefyNxl1yGzqi6\n/pTA8ef/rknAi0KIj4C5UsoNAY4prGzancP2AwUA3Hpp9zpJoMhRzGPrn67Tvn9Sn2aNT2kemqbh\nOHQQS3pndDodKTNvB4+biNQ2wQ5NCQP+dA1dWzOi51bg30IIMzBHSvn3gEfXymmaxqs1pSNMRj0X\nDUzzbVu8fzlfHD75zOCOPlPpndhTlQxohZx5ud4icbt20v6hR7B27kJEsqq2ojQfv+43pZR5eO8K\n5gJ/A54BVCL4Gb7ddow5y07O8/P0ncN9k81/n7XFlwSMeiN39ZtJj/huQYlTCRzN46Fo1UryPvoQ\nzeHA1qcvxpjwejNYCQ3+PCPQAZcBM4ExwGfA2ADH1appmlYnCcyc0IP4aAtOj4vVR9fyyX7vQKsR\nbYYwucf1ai6BVshx7BjZ82ZTtX8f+shIUm6dhn34CHXHpwSFP3cEGcAeYB4wS0pZHtiQWr8TU0sC\nvPb7UWzO/pHVGQf5eO9SXDUF5GIi7CoJtGLF36yiav8+ogYPJXnyLRhjYoIdkhLG/EkEI6WUhwId\nSLjILqjgmXe3AJDewcRb2+ezI7/uVNATOo1jTPsLVBJoZaqPH8OU2gadTkfixOuw9exF1ICBwQ5L\nUU6fCIQQw6WU3wHdhRDdT90upfwioJG1QiUV1Tz0+nc1SxpZqUvIqnmPekSbIfRMEMSaY+gco8aL\ntyae6mrylyymcMVyUqZOJ+aC0egtVpUElJDR2B3BncB3wKMNbNMAlQjO0pNzT9bSG3JZJttrinpP\n7XkTQ1IGYtCfe3ltJTRV7JFkz5uDMzsLU2KSmi1MCUmNzVk8q+a/F5yujeIfp8vD8wu3kl/inUns\nqTuH8+GRBYA3CQxvMziY4SkB4KmqJPe/iyhe9TXodMSOu5TEidejN5uDHZqi1HPGTmghxDf+rFNO\nb/mGw75qolazAaO1il0F3iodA5P7BTM0JUDKfthC8aqviWjblvYP/pHkm6eoJKCELH8eFteZzUAI\noQfU/e1Z+OTbgwBcd2FnRg9O5G8bXwbAqDNgNkQEMzSlCbnLytAZDegtVuzDR6C53diHnY/epKYP\nVUJbYw+L7wd+B8QLIWpPKBMJfBjowFqLddu9s4ehd9NNuPnbxpcpdHjvDqb1mhTEyJSmomkaZZs3\nkvPufOxDhpA8ZSo6nY6YURcGOzRF8UtjdwSzgcXAv4C7a60vkVLmBjSqViK7sII3l+7CkHCMiC7b\nePnkHPTM6j2FQSkDghec0iRcRUVkv/s25T9sQWcyYUxIPPNOihJiGntYXAgU4i0/rZyDh/7jHSoa\n0WWbb92lHcfQLqotg1L6BysspQlomkbJ/74ld+F7eCorsXYXpEyfSURKarBDU5Sz1ljX0Fwp5Qwh\nxHrqTzqPlHJEA7spQEFJFb97dR0AhsQM3/qXL3paDRFtJaqPHSN73hz0ZjPJNe8G6PTqBUClZWqs\na+jVmv8+0hyBtBaapvmSAIC58w40YGTbYSoJtHCax4OnvByD3Y45LY2UGbOw9eyFKT4h2KEpys/S\nWNfQ9zX/XXlinRDCCMTWVCNVGrB+R5bv8x3Xd2D+Ue/N1A3drg5WSEoTcGRmkj3vLXQGI+1+/yA6\nvZ6YkeoVG6V18Oc9gvlCiBghhBXYDuwXQvwm8KG1TCXlTgD6d0ng3aNvAJAW1YYIgxpC2BJpLhf5\nSxZz+InHqDpwAGNcHJrTGeywFKVJ+fMeQR8pZbEQ4npgDfBbYD3wYkAja6E+WLUP0EjrVsyeAu/d\ngJpovmWqOniArLmzqc7MwBAbS8qt01V9IKVV8icRnGhzIfCZlLJcCOEJYEwtlqZ5v/hNXbaxqsD7\n/kD/pD5EmSKDGZZyDjzV1WS+/CLu0hJiLhxN4g2TMNhswQ5LUQLCn0QghRDLgd7AwzVdRMopPJrG\n7X9bBToPxgRvEhiWOoiJXa8IcmTK2XBXVGCw2dBHRJAybTp6ixVbz17BDktRAsqf8W5TgbeAsTWT\n0iQAfwxoVC3Q32rmGNCZHL5103pNwh4RFayQlLPgrqgg+525HH78EdwVFQBEDRykkoASFs6YCKSU\nFcBKoIsQ4lKgXEr5WcAja2FOzDp2/nDvQ+E+CT2CGY5yFsq2beXw43+k+JvV6K1W3CUlwQ5JUZqV\nP3MWjwMW4B0xpAN6CSEmSym/DnRwLcXuw4W+z9a4YjgO6WpymZDnKi0h9/0FlG74DgwGEq6+lvgJ\nV6Iz+tNjqiithz//xz+Nt1toO4AQojfe+YtVEf0a//zoJwASos2UOb3dCu2i2gYzJMUPWW++TsWO\n7VjSO5MyYxbmtHbBDklRgsKfZwQRJ5IAgJRyB/4lkLBR6XAB8NvJvdiWtwOATtEdghmSchqe6mrf\n58TrbyTxxkm0f+gRlQSUsOZPIsgTQtx6YkEIcQuQH7iQWhany13zycNff3jat96kXiALKZqmUbRm\nNQf/cD+OzEwALB06En/ZeFUjSAl7/vxmfxewQAjxH7zF53YBkwMaVQtSWOodJWRIOllc7rfn3aUm\nnAkh1Tk5ZL89h8rdu9BbrThzczCnpQU7LEUJGWdMBFLKPcBgIURszXJRwKNqQR78z3fo7QVEpO8E\nYEL6JXSNTQ9yVAp4i8QVfrmC/MUfo1VXE9l/AMm3TscUFxfs0BQlpDRWhrod8BwggC3AA1LKs+oS\nEkJcDrwEGIA3pZTPnKbdDcAiYIiUctPZnCOYjpXkYe69Dn3kyeGGl3YcE8SIlNrylyymYMliDHY7\nyTNuI2rIUHQ6XbDDUpSQ09gdwRvAHuBd4Dq8SWGmvwcWQhiAV4BLgAxgoxDiUynlzlPa2YF7gQ1n\nF3rw/XXTs+hrqkek2JL4w5BfY9Kr5+jBpLlcvlIfcRdfgrukhMRrr8Ngt59hT0UJX419a7WTUo4H\nEEIsAzae5bGHAvuklAdqjvE+cA2w85R2TwLP4p0fuUWocjlYemCFb7lD4ZX8YayanzbYKg8cIHve\nbPS33Azd+2CIiiJl6vRgh6UoIa+xROCrtSuldAshzvbYacDRWssZwLDaDYQQA4H2UsqlQgi/EkFc\nnA2j8dwneElK+vm/GX66+ztWZawFwF2cwO9uvIikhNAtLNcU1xzK3A4HR959j2NLPgOPh/KDB+k4\n8vxgh9XsWvvPuSHqmptGY4mguxBi3emW/ZiqsqHOWN+Ul0IIPfAPYIYfcfoUFlacTfM6kpLs5OaW\nnvP+J3yxZw0AruwOJJYPwuDxNMlxA6GprjlUVezeRfa82ThzczElp5AyfSYdRw1p1dfckNb+c26I\nuuaz3/d0GksE15zT2U7KANrXWm4HHKu1bAf6AKtr7jZSgU+FEFeH8gNjp9tJTqV3gjbX8XRm3NQz\nyBGFr/Id28n8x3Og0xF3+QQSrr4WfYQatqsoZ6uxqSpXnm6bnzYC3YQQ6UAmcDMwpdbxi4HEE8tC\niNXA70I5CQCUOssA0Dw6tGorXdJighxR+NE8HnR6PbaevYgeMYrYsRdj6aSG7CrKuQrYK5VSShdw\nD7AC70toH0gpdwghnhBCtNgJfD/a5y286imP4aKBaejVcMRm4yop4fh/XiX/048B0On1pM66XSUB\nRfmZAjrWUUq5DFh2yrrHTtP2okDG0lT0mjd3uo51Ia6/OcjRhAdN0yjdsJ6c9xfgKSvD2q27765A\nUZSfTw16P0ubZA66ePBU2LlogKowGmjOgnxy5r9N+bYf0UVEkHTzFGLHjlNJQFGakF+JQAgxGugp\npfy3ECIZsEsp9wc2tNAz7/PduD0aRuDOq3tht6kHk4HkKi7m8OOP4KmsxNazNynTZmBKSgp2WIrS\n6vgzMc3vgIlAMvBvwALMBS4IaGQhZvuBfL7Zmol1qHc+4h4dVL2aQNE0DZ1OhzEmhphRFxKRlkb0\nyAtUeQhFCRB/5yy+CCgDkFIeAWIDGFPIcXs8vPDBj0R03+xbZzGo5wNNTXO7KVi+jOP/fsVXJiJp\n0mRiRl2okoCiBJA/XUOVUkrnKW8WewIUT0j6eM1BdNYSDLHe9wcmdBqHxWgJclSti+PoEbLmzsZx\n+BAGezSuwgJM8QnBDktRwoI/iSBDCDEc0IQQOuAPeIeDho3l237C0tf7UrXVaGFC+iVBjqj18Did\nFHz2KQXLl4HbTfT5I0maNBlDVFSwQ1OUsOFPIrgXmI/3LeAK4Du8L4eFBU3TMMTm+Jbv7n+76qZo\nIprHQ8azT1N18ADG+ARSpk0nsk+/YIelKGHHn4lpjgFja8pF62veCA4bn/7vkO/zvQPuJD1GzUXc\nVHR6PdEjRmLulE7S9Tegt1iDHZKihCV/Rg1desoyAFLKLwIUU0jZsicbU6d9wQ6j1SjfuYPCz5fR\n9u570ZvNxI65ONghKUrY86dr6NFany1AX2ArEBaJoNRw3Pe5Y3S7IEbSsrkrysn94H1K1n4Lej2V\neySRfVU3kKKEAn+6huq8LyCE6Av8KmARhZDDWaVU2PdhAPok9FAjhc5R2Q+byZ7/Du7iIsztO5Ay\nYxaWjp2CHZaiKDXOusSElPInIUT/QAQTahat3ocuqhKAqb0mBTmalil30UIKVyxHZzSSeN0NxF16\nOTqjqmyiKKHkbJ8R6IEh/uzX0lW6qjhsWYXeVkbnyO5EmUJ3BrJQFtl/AFUH9pMybQYRbVRtJkUJ\nRWf7jMAF7AduCkw4oSGj9BhPb3wRor3LA9t0D25ALYgzP4/che+RdNPNmBKTsHUXWB94SA25VZQQ\n1mgiqJlO8q9Sys+bKZ6Q8IH81Pe5c/UYxrQPq7JK50TzeChe/TW5//0QzVFFRNu2JF57PYBKAooS\n4hqtNSSl9AANzh/Qmh0v8M5CVrVtFJMGqWJnZ1KddZyMvz9DzoL56AwGUmbeTsI11wU7LEVR/ORP\n19AWIcQgKeXmMzdt+fYW7qfC6H2T+PcTL6Bdkip10JjSzZvIeuPfaC4XUYMGkzzlVowxYVWTUFFa\nPH8SwXDgF0KIXdRUIAWQUo4IWFRB9Mb2dwDwOCz06KhKTZ+JJb0zxoREEq+7HvugIcEOR1GUc+BP\nIvhDwKMIETvzJeXOCgC03RfA+CAHFII8zmoKlnyKVfQgsncfTPHxdHryKTVjmKK0YKdNBEKIt6SU\nt0kpVzZnQMH06o+zAfBURNE+Ud0NnKpy716y5r2FMysL6/59RPbuA6CSgKK0cI3dEQxstihCgMvj\nQsM7GYpj+wjsXdU0lCd4qirJ++hDilZ9DUDsxZeQOPH6IEelKEpTafUvhvlrV8EeADxVVkDPVSM7\nBTWeUFGddZyMF57DVZBPRGobUqbPwtqtW7DDUhSlCTWWCPoKIXIaWK8DNCllcoBiCop3d38IgKfE\nOytWepvoYIYTMowJiehtNuLPH0H8lVehN6k7JUVpbRpLBHuACc0VSLB5PN5uIedRwR1X9gpyNMFV\nunkj7rJyYkdfhN5kouMjj6v6QIrSijX2r9shpTzcbJEEicvj4puMdZS7ytGqzZgNFs7vkxrssILC\nVVREzoJ3KNuyGb0tkuhhw9FbLCoJKEor19i/8OpmiyKIZOF+Ptq3FADNZWJ0//ArjKZpGiX/W0vu\nB+/hqajA2q07KdNnobeostuKEg5OmwiklMObM5Bgqah2AOAuTqB673nc/EB4PQj1VFdz7F8vUbFz\nBzqzheRbphIzeowaEqooYSTs7/k3y1wA3EVJvHjP6CBH0/z0ERHorVZsffqSMnUGpoSEYIekKEoz\nC/tEcKT4GNhhaM9kom3hMSKm+vgxSrdsJuGKqwBInXUHuogIVVxPUcJUWCeCw1mlFNt/AiApuvVP\nPKO5XBSsWE7BksVoLheRvftg6ZSO3mwOdmiKogRRQBOBEOJy4CXAALwppXzmlO33AbfjnfAmF5jV\nXCOVNstcXvtyDeaakaJX9biwOU4bNFWHD5E99y0cR49iiIkl5dapWDqlBzssRVFCQMASgRDCALwC\nXAJkABuFEJ9KKXfWavYDMFhKWSGEuAt4FmiWyYFfWboR68ANAKTbO2HQG5rjtEGRt/hjCj5bAh4P\n0RdcSNKNkzDYWv8dkKIo/gnkHcFQYJ+U8gCAEOJ94BrAlwiklKtqtf8OuDWA8dRhtJf4Ps/sM7m5\nThsUOoMBU3wCKdNnYusZ3i/LKYpSXyATQRpwtNZyBjCskfa3AcvPdNC4OBtG47n/9p6UZKe4zOF7\nk3j6gBvo0aHDOR8vFLkqKji+dBlp110LQPepk9AmX48hjN4LSEqyBzuEZqeuOTwE4poDmQgaGoKi\nNdRQCHErMBg44/jNwsKKcw4oKclObm4pP+zN9a0rK3eQm1t6zscMNeU/bSP7nbm4Cgqo0gx0mzSR\n/MJK78ZSZ3CDayYnfs7hRF1zePg519xYAglkIsgA2tdabgccO7WREGIc8EdgtJTSEcB4fE7cDbQm\n7rIychYuoHT9OjAYiL/yaqIvCL/3IhRFOXuBTAQbgW5CiHQgE7gZmFK7gRBiIPAf4HIpZUOVTgMi\np6iyuU7VLMq2/kD2vNm4S0sxd0ondfoszO3bn3lHRVEUApgIpJQuIcQ9wAq8w0dnSyl3CCGeADZJ\nKT8F/g5EAYuEEABHpJRXByqmEyodbgzxWYE+TbPyVFWReOMk4sZdis7QekdAKYrS9AL6HoGUchmw\n7JR1j9X6PC6Q5z+d3TkHMSYeByDK1PKGUZ4oEhfZty/GmFiiBgwk/Zm/Y4yJDXZoiqK0QGH5ZnGG\nZb3vSfaQlJY1I2d1bg7Z8+ZQuXsX9vNH0Oa2OwFUElAU5ZyFXSJwup3obN53CH436J4WU19H83go\nWvkleR//F626msh+/UmceEOww1IUpRUIu0Tw/s7PANDcetJjWsb7A9VZWWTNfp2qAwfQR0WRMn0m\n9qHDW0wSUxQltIVdIsgr9Y7BteT3CXIkZ0EHjqNHsQ8dTtLkKRjtaj5lRVGaTtglgpJyJxigd2L3\nYIfSqKpDBwEdlk6diEhJpdMTT2FKSgp2WIqitEJhNw3VsbxyANonRwU5koZ5HA5yF73Pkb8+Qdbs\nN9A8HgCVBBRFCZiwuiPILjhZnqJLWkwQI2lYxe5dZM+bgzM3B1NSMslTblVTRiqKEnBhlQjyar1R\nHGkJnUv3VFWS+8FCitesBp2OuEsvJ+GaiWrCGEVRmkXofBs2g9yQLS2ho3zHT0SktSNl+iysnTsH\nOyBFUcJIWCWCUCo25yotoTojA1vPXugtFtrd93tMCYnojGH1I1EUJQSE1bfO/oyiYIeApmmUfr+B\n3PfeRXO76PTk0xhjY4lISQ12aIqihKmwSgSrt2Sg7xS8ZOAsKCBn/jzKt/2ILiKCxGuvxxCt3glQ\nWofi4iJ+/etfAlBQkI9eryc2Ng6AN96Yh8lkOuMxnnrqz9x663Q6dOjUJDE5nU6uvvoyrrvuRu64\n4y7f+okTJ/D22wux2701+jdu3MBHH33A008/D8C6dWt5881/U13twOPxMGrUaH75y3vrHX/u3DdZ\nvnwpBoOB3/72AYYMqT/31vfff8drr72MpoHNZuOPf/wTaWntcDgcPPnko+zdu4fY2Dj+/OenSU1N\nZfnypSxcuMC3/759e3j77ffp3Lkrzz33HJ98spjKygo+/3x1k/wdQZglAoPRjT7S+0KZ1WhrtvNq\nmkbxmm/I+3AhnspKrD16kjJtJhHJyc0Wg6IEWkxMLHPner/A3nrrP1itNqZMmVqnjaZpaJqG/jSj\n4R5++PEmjWnDhnV06pTOypVf1kkEjdm7dw8vv/w8zz77Ih06dMTlcrFkySf12u3bt5c1a1Yxf/4i\ncnKyuf/+X7FgwX/rXdvf//40L7zwT9q378CiRe/zzjtzePDBR/n004+Jj09g4cJPWLFiGf/5z794\n/PG/MH78lYwffyUAe/bs5rHHHqZz564AXHzxxYwffy3TpjXt1O5hlQhKnCVE1HyOMTfvFHdlmzYC\nkDJtJtEXXKjKQygB9cHX+9i4u2mn+BjSI5mbxnY96/0yMo7y0EP306/fAHbu3M6zz77I7NlvsGfP\nbhwOBxdffAkzZ94BwF133cZ99z1AenoXrrxyHNdccz3ffbcOi8XCM888T1xc/Fmd+8svV3Dzzbew\ncOECdu3aQc+evc+4z7vvzmP69Nvo0KEjAEajkYkN1PVau/Ybxo27DJPJRFpaO1JSUpFyV71z6HRQ\nXu59f6m8vIzExCTf/v/3f/cAMHbsJbz88vP1zvHVVyu45JLLfMsDBw7k+PFCP6/ef2E1SP3Ew+JR\nacMDfi7N7aZ85w4AdDodKTNn0fGJp4i5cLRKAkrYOXToIFdeeQ1z5iwgKSmZu+66h7feeoe5cxew\nceMGDh48UG+fsrIyBgw4j3nz3qNPn74sXfrpWZ2zsrKSrVu3MHz4SMaNu4yvvlrh134HD+5HiJ4N\nbvvmm1XMmfMGALm5OSQnp/i2JSUlk5ubW2+fBx98lPvuu4eJEyewcuUXTJkyrd7+JpMJi8VKaenJ\naSg1TePrr79i3LjL6h2zqYXNHYHL7Wm2czkyjpI1dzaOQwdp98BD2LoLTPEJzXZ+RblpbNdz+u09\nUNLS2tX5TfnLL1fw2WeLcbvd5OXlcujQAdLT6w6bNpvNnH/+SACE6MmPP/5wVudcu/YbhgwZhtls\nZsyYcdx++1Tuvvs36PX60/wyduZf0EaPHsPo0WMA7xd1vSM0cIgPPljACy/8kx49evHOO3N45ZUX\n+f3vHz7N/icP8NNPP2K32+nYsdMZ4/q5wiYRlFcGfuJ2j9NJwbKlFCxbCm439uHnY27TNuDnVZRQ\nZ7FYfZ+PHj3CokXv88Yb87Db7TzxxKNUV1fX26f2w2W9Xo/b7a7X5te//iXFxUX07t2H3//+4Trb\nvvpqBTt37uCGG64CoLCwkK1bt3DeeYOJjo6htLTE97C4pKSE2FjvnB6dOnVGyl107tyl0WtKTk4h\nJyfbt5ybm+Pr9jkhLy+Pw4cP0aNHL8DbBfTww7+rs39CQiJOp5Oqqkqiok6Wvlm58otmuRuAMOsa\nCqTKA/s58uSfKFiyGGN0DG3v/Q1tbv8FBnvzPotQlFBXXl6OzWYjMjKSvLw8vv9+/Tkf66WXXmXu\n3AX1kkBJSQk7d+7g44+X8eGHS/jwwyX85je/83UPDRw4iBUrvJMnulwuvvxyOQMHDgbgllumMW/e\nbDIyjgLgdrt5//359c49cuSFfPXVCpxOJ5mZGWRlHa/XpRQTE0NxcbHvWBs3bqBjx3Tf/suXLwXg\n66+/ZMiQk13Wbreb1atXcvHFzZMIwuaOINDKNm+k+lgmMReNJfH6GzFYrWfeSVHCkBA9SE9PZ9q0\nSbRtm0bfvv2b/BzffPM1Q4YMw1jrBc0LLriI119/lfvu+wOzZt3Jc889xfTpkwGN4cNH+h7Kdu/e\ng3vu+TWPPfYgDocDnU7HqFGja467igMH9jFz5h1069adCy64iFtuuQGDwcD99z/oGzF033338Oij\nTxAXF88DDzzMQw/dj15vIDo62jcy6pprJvLEE48yadK1xMTE8uc/P+WLdcuWTbRt247U1LrvFz39\n9NN89tkyysvLmThxAtdccx0zZtz+s/++dA31U4Wy3NzScwq4uMzBfW+uwNJvLaPShjNZXPezY6k8\nsB9Lp3R0ej0ehwPHkcNYu4VWeeukJDu5uaVnbtiKqGsOD+qaz3rf0z4ECauuIZ25aWoNuSvKyZo3\nm6NPPUnx6q8B0JvNIZcEFEVR/BE2XUMOpxu9tQyAanf9B1P+KvthC9nz38ZdXEREu/ZYzvBASVEU\nJdSFTSIorXCiad47owFJfc96f1dJCbnvzad04/fojEYSrr2O+MsnqCJxiqK0eGHzLZbzM0tQV+za\nSenG77F07kLKjFmY26Y1UWSKoijBFTaJwKA/+7d5nQX56C0WDLZI7EOHoTMaiRp4npo1TFGUVkV9\nozVA83goWvU1hx/7I4JMbi0AABDESURBVLmLFgLeN/7sgwarJKAoSqsTNncER3PKMNjPXKypOiuL\n7Hmzqdy7B73NhrVLt2aITlFavqYoQw2wdOlizj9/JAkJiWcdw65dO7jjjum8+OKrDB48FPAWvXvk\nkT/4KqMCvP76q8TGxnLTTVPQNI0FC95m6dLFGI1GDAYjU6ZM5dJLx9c59unKRtd28OAB/vznR3zL\nmZkZ/N//3c31109iz57dPPfcM1RVVdK2bRqPPfYXbDYb27f/f3t3Hh9VkS1w/JcNGDRsCpHFQRA8\nI0pA2QTBBXAei4KDjMAIo4CgAo6KC+KCDuoDxIeg4zaCbC6AvPjMABrNg8hDFiMI7geRYZQliBog\nDBAS0u+PujRN1g7pDiR9vp8Pn09ud91bVZ3Q51bde099ztSpk4iKisLn8zFixO3+5xbAPfA2fPhg\nGjRoyMyZfy/1ZxKMiAkEeXFZxNRxj4PXqFLwaV/f0aNkfpDCL8nv4MvJ4cxL2lDvpiHEeo+dG2OK\nF0wa6mAsXZqMyO9OKhCkpqaQmNia1NQUfyAoSVLSIjZsWM+sWfOpXv0MsrKyWLXqowLlikobHahJ\nk6b+zyA3N5frr+9Jly5XATBp0kTuuecBEhNb8+67SSxY8DrDho2kWbMLmDVrPrGxsfz88x6GDr2J\nTp26+B9OW7jwDZo0OZ/s7MOl/jyCFTGBYFvuF/6fm9T8bYH3j2Rk8PM7i4k580zq3TqS+DbtyrN5\nxoRU0pYlfPbTFyUXLIVL6rWkX7NrT2rf995bQlLSInJycrn44kTGjn2AvLw8Jk36K999txmfz0ef\nPv2oU6cOW7ZsZsKE8VStWq1UI4m8vDzS0pbz3HMvM3r0CHJycoLad9682bzyymyqVz8DgPj4eP96\nAIGCSRsdKD19Leed18SfYXTHjh0kJrYGoF27Dowffy/Dho2kWrVq/n2ys7OB4wntMjJ2kZ6+jptu\n+jNJSW+X2JeTFTGBIOHHPUTlZdO01Q3+1/Jycjh64ABxtWtTtWFD6t82iuryO2ICEj8ZY8pm69Yt\nrFy5gpdeeo3Y2FimTHmK1NQPaNiwEXv37mPePHcdLisri/j4eBYvXsjYsQ/QvLmUqp6NGzfQuPF5\nNGzYiJYtW7Fu3eoTplgKs3//fo4ezeWcc+oX+v4rr7xAYmIrOnbsXGTa6Pgi8omlpn5A9+6/9283\nbtyY1atX0alTZ1asSGX37uMJ6774YhNTpjzF7t27eOyxJ4mJiQFgxoz/YvTou9i7N7wrK0ZEIMjL\nOULL9M3kcISDrd3CFoe+38LuOa8RXa0a545/hKjoaOLbtD3FLTUmNPo1u/akz95D7dNPP+Gbb77m\n1ltdHv7s7MMkJCTQocNl/PDDv5g+/Rk6dryc9u3Ltk5IamoK3bq5L97u3X/Phx+m0Llzcet/RAHF\nZ6y57bbR/p9LShsdKDs7mzVrPubOO8f6X3vooceZMeMZZs58mS5drjwhD1LLlq14/fVFbN36PZMm\nTaRDh06sWfMxCQkJNG8upKevK7adZRXWQCAiPYAZQAwwU1Un53u/KjAPaAP8AgxQ1W2hbsevy5Zy\nxgE3vxa9ehU/fbWJvf+bCj4fNa/uhi83l6gqVUo4ijHmZPh8Pnr37lPoUpFz577F2rWrWbx4AWlp\nyxk37uEij7Nr107Gj3cpnG+44UaGDTt+/SE3N5eVK9NYu3Y1s2e/Sl5eHgcOHODQoUP+lNOB9u/f\nz3nnNaFGjZrExMSSkZFR4MJvfiWljQ60Zs0qWrS4yJ/aGtz1g+nTXwTcReV16wpmXW3a9Hzi4uLY\ntm0rX365iY8+WsGqVSs5cuQI//73AcaNG8d99z1SYL+yCtu9kCISA7wA9ARaAINEpEW+YsOBTFVt\nBjwLTAl1O47s+YnM95f5t+PT17I39UPiEhI4d9xDJNw0hGgLAsaETdu2HVi+/EP/9Ma+fXvJyMgg\nMzMT8NG1a3eGD7+NzZu/BaB69TM4ePBggePUr9+AOXPeZM6cN7nuuutPeO+TT9Zy4YUtSEpayuLF\n/yApaSmXX96FVas+Ij4+nvj4Gnz22Xp//enpa/1ZTwcPvoVp0yZz8KBbTjIrK4vk5HcK1F9c2uj8\nUlNTCqwlkJn5K+CuZcydO4u+fV3iy507d/jXWti5cwc7dvxIQkJ9Ro26y59G+9FHJ9K+/WVMmRLy\nr0ggvCOC9sAWVd0KICILgL7A1wFl+gKPez8vBv4mIlGqGrKUqHveegNfzvFFaaKAuHr1aPzYRKLj\nLAAYE27nn9+MoUNHcvfdo/D58oiJieX++8cTHR3D5MkT8fncyl533PEXAHr1uo7Jk58o1cXi1NQU\nrrji6hNeu+qqrixd+g+uuaYHEyZMZNq0pzlwwOUbGzHiDup7i0b17z+Aw4cPMWzYEOLiYomNjfUv\nJxl4jaCotNG7d2cwbdoUpkx5FoCDBw+yYcN6f7rpY95/fxnJyUn4fHD11d3o0aM34K5tvPmmu2so\nOjqa++9/iBo1apzsx31SwpaGWkT6Az1U9VZvewjQQVXHBJT50iuz3dv+3ivzc1HHzc096ouNjQm6\nHV8/+Z9kpq8/4bXa7drQ4pGHitjDGGMqpSLTK4RzRFBYpfmjTjBlTpCZWXDIWJya/Qawd+Pn/lFB\nVFwctfoNjJg85pazPTJYnyNDGdcjKPK9cOZL2A6cG7DdCNhZVBkRiQVqAr+GshFV6tajdo9e/u06\nPXsTV7duMXsYY0xkCWcgSAeai0gTEakCDASS85VJBm72fu4PLA/l9YFj6vTqTdzZdal2TgK1e/Yq\neQdjjIkgYZsaUtVcERkDpOBuH31NVb8SkYnAp6qaDMwC5ovIFtxIYGA42hIdV4W6A/9EzVrVOWoX\niI0x5gQRs2Yx2JxipLA+Rwbrc6n3tTWLjTHGFM4CgTHGRDgLBMYYE+EsEBhjTISrcBeLjTHGhJaN\nCIwxJsJZIDDGmAhngcAYYyKcBQJjjIlwFgiMMSbCWSAwxpgIZ4HAGGMiXFgXrz9VRKQHMAOX9XSm\nqk7O935VYB7QBvgFGKCq28q7naEURJ/HArcCucAeYJiq/qvcGxpCJfU5oFx/4G2gnap+Wo5NDLlg\n+iwiN+KWgPUBm1T1T+XayBAL4m/7t8BcoJZX5kFVXVbgQBWEiLwGXAv8pKoXF/J+FO7z6AUcBG5R\n1Q1lqbPSjQhEJAZ4AegJtAAGiUiLfMWGA5mq2gx4FgjPitDlJMg+fwa0VdVE3PrQT5dvK0MryD4j\nIvHAX4B15dvC0AumzyLSHBgPXK6qFwF3l3tDQyjI3/MjwCJVvQSXyv7F8m1lyM0BehTzfk+gufdv\nJPBSWSusdIEAaA9sUdWtqnoEWAD0zVemL+4MAtyXYjcvylZUJfZZVVeo6rF1PtfiVoyryIL5PQM8\ngQt6h8uzcWESTJ9HAC+oaiaAqv5Uzm0MtWD67AOOrfZek4IrIVYoqrqS4ldq7AvMU1Wfqq4FaolI\n/bLUWRkDQUPgx4Dt7d5rhZZR1VxgH3BWubQuPILpc6DhwHthbVH4ldhnEbkEOFdVl5Rnw8IomN/z\nBcAFIvKxiKz1plUqsmD6/DgwWES2A8uAO8unaadMaf+/l6gyBoLCzuzzJ1QKpkxFEnR/RGQw0BaY\nGtYWhV+xfRaRaNy0373l1qLwC+b3HIubMrgKGATMFJFaYW5XOAXT50HAHFVthJs3n+/9/iurkH9/\nVcYPaztwbsB2IwoOFf1lRCQWN5wsbih2ugumz4hId+BhoI+qZpdT28KlpD7HAxcDaSKyDbgMSBaR\ntuXVwDAI9m/7XVXNUdV/AooLDBVVMH0eDiwCUNU1QDXg7HJp3akR1P/30qiMdw2lA81FpAmwA3fx\nKP9dE8nAzcAaoD+wXFUr8oigxD570ySvAD0qwbwxlNBnVd1HwJeBiKQB91Xwu4aC+dv+H7wzZBE5\nGzdVtLVcWxlawfT5B6Abrs8X4gLBnnJtZflKBsaIyAKgA7BPVXeV5YCVbkTgzfmPAVKAb3B3E3wl\nIhNFpI9XbBZwlohsAcYCD56a1oZGkH2eCpwJvC0iG0Uk+RQ1NySC7HOlEmSfU4BfRORrYAVwv6r+\ncmpaXHZB9vleYISIbALewt1OWWFP7ETkLdxJqojIdhEZLiK3i8jtXpFluOC+BXgVGFXWOm09AmOM\niXCVbkRgjDGmdCwQGGNMhLNAYIwxEc4CgTHGRDgLBMYYE+Eq43MEpgLxHvY6zPFcQCtU9Z4S9tkO\ndFfVb0NQ/5O4/Dy7gKrAKmCUquacxLFGAzGq+pyIXAo0VdXF3nsxwHqgvZczp8y8zyELOALEAVNV\ndXYQ+/UDfqjgz1SYELJAYE4H/VX1y1NY/2xVfVBEqgErcYGh1BksVfWFgM1Lge64pIao6lGgdQja\nmt8fVPVbEWkFpIvIMlXdXcI+/XABzwKBASwQmNOUiAzBPUhUBZdHZayqphVSbiJwI5ANHAWuVNUs\nEekITMI9RAfwqKoWm2hPVQ+LyCpAvGP3Bp7E5bjfDdymqlu9p1dnA7/heI786d7oIhaYBkwA4kVk\nI96DXUCOt88goJeq/tGrJw6XRKwdLn3AeOB63Fn+D8CIkp4GV9VNIpIFNAB2i0hr4G9AddxI52VV\nfV5EeuHy8VzpPaA0VVXfEJFhwO1e+zOB21X1u+LqNJWHXSMwp4PF3tPOG0XkP7zXlqlqBy/H/GDc\nQkInEJG6wF1Aa1VtBVwJHBKROrgz+gGq2haXtvdVEamR/xj5jlcLuAb4TETOwaUqHxSwhsN8r+gY\nXD6fVt7CIXMDj+N9aU8EUlS1dSFTXW8DXUWktrd9LfC5qv6IS33SCLjM63sqQSQIFJErcflmjo2s\ntgJdVfVSXJ6lMSJygbdgyzLgKa9tb4jIVbjA09krPx2YWVKdpvKwEYE5HRQ2NdTce9S+AW5VtYYi\ncraq/hxQJhP3hTdfRFKAJd5ooDPQBEgRkWNlfUBTYGMh9Q/10jX7gHdwQacv8GnAdYhZwPMiUh03\nfTRJRH6DO9tPK01nVfWAiCzBjQxeBG7BjTAA+uCmkDZ4bY/FraJXlHe86w9NcckEj13bOAN4WURa\nAnnAOUAisLmQY/TBTWV94tUZhUvaZyKEBQJzuloIjFbVJd4X3SFcMjE/Vc0VkXZAZ6Ar7ky+O+6L\nbIOqdg2yrtmqekK+KW+hokLzr6jqQhH5GDd6eBh3Fn9L0D1z5gCTReRtoBMwwHs9CnhcVQuMgIpw\n7BrBIOBNEWmuqnuAybhppSGqelRElpPv8wsQBfxdVSeWsg+mkrCpIXO6qgn80/t5JG6+/ATeVM/Z\nqpqmqhNwSckuAj4GWojIFQFl25ey/tVAWxG5wNseCnyiqge95SB3enfoPIFbRSu//V4fipIG1AWe\nAv5bVY/dNZUMjD62hoCIVBORxJIaq6pv4UYnD3gv1QJ+9IJAK+DyYtqWDNwsIg28OmNEpE1JdZrK\nw0YE5nR1N7DEu0VyBW4VufxqA4u86Zoo3F0w76pqtoj0BZ72vlDjcFNI1wZbuapmiMgtwEJvkZM9\nwJ+9twcCA0XkCG7UUNi6wB8C93gZMZfjLhYHHt8nIvOAx4COAa/PFpGzgJXeNE008DzweRDNfhBY\nJyJTcdco5onIzbgslf8XUG4e8JqIDASe8a4T/BVY6vU1DjciWx9EnaYSsOyjxhgT4WxqyBhjIpwF\nAmOMiXAWCIwxJsJZIDDGmAhngcAYYyKcBQJjjIlwFgiMMSbC/T+bhiiOkCYUbgAAAABJRU5ErkJg\ngg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "lin_svc_ = SVC(C=0.01, kernel='linear', probability=True)\n",
    "lin_svc_.fit(X_train,y_train)\n",
    "\n",
    "# 预测\n",
    "y_train_pred = lin_svc_.predict(X_train)\n",
    "y_test_pred = lin_svc_.predict(X_test)\n",
    "# 预测概率\n",
    "y_train_proba = lin_svc_.predict_proba(X_train)[:, 1]\n",
    "y_test_proba = lin_svc_.predict_proba(X_test)[:, 1]\n",
    "# 模型评估\n",
    "model_metrics(X_train, y_train, X_test, y_test, y_train_pred, y_test_pred, y_train_proba, y_test_proba)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## DT模型"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 318,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "最优分数： 0.729497046466\n",
      "最优参数： {'min_samples_split': 72, 'max_depth': 5, 'class_weight': 'balanced'}\n"
     ]
    }
   ],
   "source": [
    "dt_param = {\n",
    "    'max_depth': [m for m in range(3,8)], \n",
    "    'min_samples_split': range(2, 100, 5), \n",
    "    'class_weight': ['balanced',None]\n",
    "}\n",
    "dt_grid = GridSearchCV(dt, dt_param, cv=n_fold, scoring=scoring)\n",
    "dt_grid.fit(X_train, y_train)\n",
    "print '最优分数：', dt_grid.best_score_ # 最优分数\n",
    "print '最优参数：', dt_grid.best_params_ # 最优参数\n",
    "# print '交叉验证结果：', dt_grid.cv_results_ # 结果"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## XGBoost模型"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "xgb_grid = GridSearchCV(xgb, xgb_param, cv=n_fold, scoring=scoring)\n",
    "xgb_grid.fit(X_train, y_train)\n",
    "print '最优分数：', xgb_grid.best_score_ # 最优分数\n",
    "print '最优参数：', xgb_grid.best_params_ # 最优参数\n",
    "# print '交叉验证结果：', xgb_grid.cv_results_ # 结果"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## LightGBM模型"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "lgb_grid = GridSearchCV(lgb, lgb_param, cv=n_fold, scoring=scoring)\n",
    "lgb_grid.fit(X_train, y_train)\n",
    "print '最优分数：', lgb_grid.best_score_ # 最优分数\n",
    "print '最优参数：', lgb_grid.best_params_ # 最优参数\n",
    "# print '交叉验证结果：', lgb_grid.cv_results_ # 结果"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 模型融合"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 参考\n",
    "\n",
    "[stacking代码参考](https://github.com/emanuele/kaggle_pbr/blob/master/blend.py#L76)\n",
    "\n",
    "[StackingCVClassifier](http://rasbt.github.io/mlxtend/user_guide/classifier/StackingCVClassifier/)\n",
    "\n",
    "[StackingClassifier](https://www.jianshu.com/p/1df64dee0c05)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 自己写接口"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def blend(clfs, X, y, X_submission):\n",
    "    from sklearn.model_selection import StratifiedKFold\n",
    "    \n",
    "    dataset_blend_train = np.zeros((X.shape[0], len(clfs)))\n",
    "    dataset_blend_test = np.zeros((X_submission.shape[0], len(clfs)))\n",
    "    \n",
    "    n_folds = 5\n",
    "    # skf = list(StratifiedKFold(y, n_folds))\n",
    "    skf = StratifiedKFold(n_folds)\n",
    "    skf_data = []\n",
    "    \n",
    "    for train, test in skf.split(X, y):\n",
    "        skf_data_temp = [train, test]\n",
    "        skf_data.append(skf_data_temp)\n",
    "    \n",
    "    print \"Creating train and test sets for blending.\"\n",
    "    \n",
    "    for j, clf in enumerate(clfs):\n",
    "        print j, clf\n",
    "        dataset_blend_test_j = np.zeros((X_submission.shape[0], n_folds))\n",
    "        for i, (train, test) in enumerate(skf_data):\n",
    "            print \"Fold\", i\n",
    "            X_train = X[train]\n",
    "            y_train = y[train]\n",
    "            X_test = X[test]\n",
    "            y_test = y[test]\n",
    "            \n",
    "            clf.fit(X_train, y_train)\n",
    "            y_submission = clf.predict_proba(X_test)[:, 1]\n",
    "            dataset_blend_train[test, j] = y_submission\n",
    "            dataset_blend_test_j[:, i] = clf.predict_proba(X_submission)[:, 1]\n",
    "        dataset_blend_test[:, j] = dataset_blend_test_j.mean(1)\n",
    "    \n",
    "    print \"Blending.\"\n",
    "    clf = LogisticRegression()\n",
    "    clf.fit(dataset_blend_train, y)\n",
    "    y_submission_proba = clf.predict_proba(dataset_blend_test)[:, 1]\n",
    "    y_submission = clf.predict(dataset_blend_test)\n",
    "    return y_submission_proba, y_submission"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Creating train and test sets for blending.\n",
      "0 LogisticRegression(C=0.05, class_weight=None, dual=False, fit_intercept=True,\n",
      "          intercept_scaling=1, max_iter=100, multi_class='warn',\n",
      "          n_jobs=None, penalty='l1', random_state=None, solver='warn',\n",
      "          tol=0.0001, verbose=0, warm_start=False)\n",
      "Fold 0\n",
      "Fold 1\n",
      "Fold 2\n",
      "Fold 3\n",
      "Fold 4\n",
      "1 SVC(C=0.01, cache_size=200, class_weight=None, coef0=0.0,\n",
      "  decision_function_shape='ovr', degree=3, gamma=0.01, kernel='rbf',\n",
      "  max_iter=-1, probability=True, random_state=None, shrinking=True,\n",
      "  tol=0.001, verbose=False)\n",
      "Fold 0\n",
      "Fold 1\n",
      "Fold 2\n",
      "Fold 3\n",
      "Fold 4\n",
      "2 SVC(C=0.01, cache_size=200, class_weight=None, coef0=0.0,\n",
      "  decision_function_shape='ovr', degree=3, gamma='auto_deprecated',\n",
      "  kernel='linear', max_iter=-1, probability=True, random_state=None,\n",
      "  shrinking=True, tol=0.001, verbose=False)\n",
      "Fold 0\n",
      "Fold 1\n",
      "Fold 2\n",
      "Fold 3\n",
      "Fold 4\n",
      "3 DecisionTreeClassifier(class_weight='balanced', criterion='gini', max_depth=5,\n",
      "            max_features=None, max_leaf_nodes=None,\n",
      "            min_impurity_decrease=0.0, min_impurity_split=None,\n",
      "            min_samples_leaf=1, min_samples_split=72,\n",
      "            min_weight_fraction_leaf=0.0, presort=False, random_state=None,\n",
      "            splitter='best')\n",
      "Fold 0\n",
      "Fold 1\n",
      "Fold 2\n",
      "Fold 3\n",
      "Fold 4\n",
      "4 XGBClassifier(base_score=0.5, booster='gbtree', colsample_bylevel=1,\n",
      "       colsample_bytree=1, gamma=0, learning_rate=0.1, max_delta_step=0,\n",
      "       max_depth=3, min_child_weight=1, missing=None, n_estimators=100,\n",
      "       n_jobs=1, nthread=None, objective='binary:logistic', random_state=0,\n",
      "       reg_alpha=0, reg_lambda=1, scale_pos_weight=1, seed=None,\n",
      "       silent=True, subsample=1)\n",
      "Fold 0\n",
      "Fold 1\n",
      "Fold 2\n",
      "Fold 3\n",
      "Fold 4\n",
      "Blending.\n"
     ]
    }
   ],
   "source": [
    "clfs = [LogisticRegression(C = 0.05, penalty = 'l1'), \n",
    "        SVC(C=0.01, kernel='rbf', gamma=0.01, probability=True), \n",
    "        SVC(C=0.01, kernel='linear', probability=True), \n",
    "        DecisionTreeClassifier(class_weight='balanced', max_depth=5, min_samples_split=72,), \n",
    "        XGBClassifier()]\n",
    "y_submission_proba, y_submission = blend(clfs, X_train, y_train, X_test)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "融合之后，指标略有提升。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 112,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[准确率]\n",
      "测试集：0.8045\n",
      "[精准率]\n",
      "测试集：0.7090\n",
      "[召回率]\n",
      "测试集：0.3743\n",
      "[f1 score]\n",
      "测试集：0.4899\n",
      "[auc]\n",
      "测试集：0.8015\n"
     ]
    }
   ],
   "source": [
    "from sklearn.metrics import accuracy_score, precision_score, recall_score, f1_score, roc_auc_score, auc, roc_curve\n",
    "\n",
    "print '[准确率]'\n",
    "print '测试集：{:.4f}'.format(accuracy_score(y_test, y_submission))\n",
    "print '[精准率]'\n",
    "print '测试集：{:.4f}'.format(precision_score(y_test, y_submission))\n",
    "print '[召回率]'\n",
    "print '测试集：{:.4f}'.format(recall_score(y_test, y_submission))\n",
    "print '[f1 score]'\n",
    "print '测试集：{:.4f}'.format(f1_score(y_test, y_submission))\n",
    "print '[auc]'\n",
    "print '测试集：{:.4f}'.format(roc_auc_score(y_test, y_submission_proba))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 采用已有的接口"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 5 classifiers...\n",
      "Fitting classifier1: logisticregression (1/5)\n",
      "LogisticRegression(C=0.05, class_weight=None, dual=False, fit_intercept=True,\n",
      "          intercept_scaling=1, max_iter=100, multi_class='warn',\n",
      "          n_jobs=None, penalty='l1', random_state=None, solver='warn',\n",
      "          tol=0.0001, verbose=1, warm_start=False)\n",
      "[LibLinear]Fitting classifier2: svc (2/5)\n",
      "SVC(C=0.01, cache_size=200, class_weight=None, coef0=0.0,\n",
      "  decision_function_shape='ovr', degree=3, gamma=0.01, kernel='rbf',\n",
      "  max_iter=-1, probability=True, random_state=None, shrinking=True,\n",
      "  tol=0.001, verbose=1)\n",
      "[LibSVM]Fitting classifier3: svc (3/5)\n",
      "SVC(C=0.01, cache_size=200, class_weight=None, coef0=0.0,\n",
      "  decision_function_shape='ovr', degree=3, gamma='auto_deprecated',\n",
      "  kernel='linear', max_iter=-1, probability=True, random_state=None,\n",
      "  shrinking=True, tol=0.001, verbose=1)\n",
      "[LibSVM]Fitting classifier4: decisiontreeclassifier (4/5)\n",
      "DecisionTreeClassifier(class_weight='balanced', criterion='gini', max_depth=5,\n",
      "            max_features=None, max_leaf_nodes=None,\n",
      "            min_impurity_decrease=0.0, min_impurity_split=None,\n",
      "            min_samples_leaf=1, min_samples_split=72,\n",
      "            min_weight_fraction_leaf=0.0, presort=False, random_state=None,\n",
      "            splitter='best')\n",
      "Fitting classifier5: xgbclassifier (5/5)\n",
      "XGBClassifier(base_score=0.5, booster='gbtree', colsample_bylevel=1,\n",
      "       colsample_bytree=1, gamma=0, learning_rate=0.1, max_delta_step=0,\n",
      "       max_depth=3, min_child_weight=1, missing=None, n_estimators=100,\n",
      "       n_jobs=1, nthread=None, objective='binary:logistic', random_state=0,\n",
      "       reg_alpha=0, reg_lambda=1, scale_pos_weight=1, seed=None,\n",
      "       silent=True, subsample=1)\n",
      "[准确率]\n",
      "训练集：0.8687\n",
      "测试集：0.7765\n",
      "[精准率]\n",
      "训练集：0.8139\n",
      "测试集：0.5661\n",
      "[召回率]\n",
      "训练集：0.6180\n",
      "测试集：0.4665\n",
      "[f1 score]\n",
      "训练集：0.7025\n",
      "测试集：0.5115\n",
      "[auc]\n",
      "训练集：0.9128\n",
      "测试集：0.7605\n",
      "[roc曲线]\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEWCAYAAACJ0YulAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4yLCBo\ndHRwOi8vbWF0cGxvdGxpYi5vcmcvhp/UCwAAIABJREFUeJzt3Xd4VGX2wPHvSSihhC41hFBFigQS\nQBAXFFBkFwiCgK4NC4sdXX+rrhqKZUVUFMFVVxEbxYagNMGGSkd6FI0IJAjSOyHt/P6YyTiElEnI\nZNr5PA+PmZk7954bcM687byiqhhjjDEAYb4OwBhjjP+wpGCMMcbFkoIxxhgXSwrGGGNcLCkYY4xx\nsaRgjDHGxZKCMcYYF0sKJuiIyHYROSUix0Vkj4hME5HKuY7pKiJfisgxETkiIp+KSKtcx1QRkRdE\nZKfzXMnOx7VK946MKT2WFEyw6qeqlYFYoD3wcM4LItIF+ByYA9QHGgMbgO9FpInzmHLAF0BroA9Q\nBegKHAA6eStoESnjrXMb4wlLCiaoqeoeYBGO5JDjGeBtVX1RVY+p6kFVfRRYAYxxHnMDEA0MVNUk\nVc1W1b2q+riqzs/rWiLSWkQWi8hBEflDRP7tfH6aiDzhdlwPEUl1e7xdRB4UkY3ACRF5VEQ+zHXu\nF0VkkvPnqiLyhojsFpFdIvKEiISf46/KGMCSgglyIhIFXAkkOx9XxPGN/4M8Dn8f6O38uRewUFWP\ne3idSGAJsBBH66MZjpaGp64B/gpUA94B+opIFee5w4EhwHTnsW8Bmc5rtAcuB24twrWMyZclBROs\nPhGRY0AKsBcY7Xy+Bo5/97vzeM9uIGe8oGY+x+Tnb8AeVX1OVdOcLZCVRXj/JFVNUdVTqroD+AFI\ncL52GXBSVVeISB0cSW6Uqp5Q1b3ARGBYEa5lTL4sKZhglaCqkUAPoCV/ftgfArKBenm8px6w3/nz\ngXyOyU9D4NdiReqQkuvxdBytB4Br+bOV0AgoC+wWkcMichh4Fah9Dtc2xsWSgglqqvoNMA141vn4\nBLAcuDqPw4fwZ5fPEuAKEank4aVSgKb5vHYCqOj2uG5eoeZ6/AHQw9n9NZA/k0IKcBqoparVnH+q\nqGprD+M0pkCWFEwoeAHoLSI5g80PATeKyD0iEiki1Z0DwV2Asc5j3sHxAfyRiLQUkTARqSki/xaR\nvnlc4zOgroiMEpHyzvN2dr62HscYQQ0RqQuMKixgVd0HfA28Cfymqj86n9+NY+bUc84ps2Ei0lRE\nuhfj92LMWSwpmKDn/IB9G3jM+fg74ArgKhzjBjtwDNh2U9VfnMecxjHY/BOwGDgKrMLRDXXWWIGq\nHsMxSN0P2AP8AlzqfPkdHFNet+P4QJ/lYejTnTFMz/X8DUA5IAlHd9iHFK2ry5h8iW2yY4wxJoe1\nFIwxxrhYUjDGGONiScEYY4yLJQVjjDEuAVd8q1atWhoTE+PrMIwxJqCsXbt2v6qeV9hxAZcUYmJi\nWLNmja/DMMaYgCIiOzw5zrqPjDHGuFhSMMYY42JJwRhjjIslBWOMMS6WFIwxJgAkH0omYU4CyYeS\nvXodryUFEZkqIntFZHM+r4uITHJuhr5RRDp4KxZjjAlkJzNOcscXd7Dt8Dbu/OJOTmac9Nq1vNlS\nmIZjw/P8XAk0d/4ZAfzXi7EYY0zASlyWyMG0gyjKgVMHGL1sdOFvKiavrVNQ1aUiElPAIQNwbJ6u\nwAoRqSYi9Zz14o0x5izTV+5kzvpdvg6jVB0K/449Zb5CJR2A09mn+Trla2b/MpuBzQeW+PV8uXit\nAWduQZjqfO6spCAiI3C0JoiOji6V4IwxJaskPtBX/nYQgM6Na5RESD7V8+R8Lj71VaHH3VzvCCpn\nPpeWlcYLP7wQdElB8nguz80dVPU14DWA+Ph42wDCmGLy5TftkvhA79y4BgNiG3BtZz/9crjmTdj0\noWfH7v7O8d9G3fJ8+SjZPC+HOCkgyhmJISI8gvvi7jvHYPPmy6SQimOz8xxRwO8+isWYgHCuH+q+\n/Kbt9x/oJWHTh7BnE9RtW/ixjbpB28EQP/ysl7Kys7h+7lVsP3qK4a2Hk3I0hW93fcvprNOUDytP\nj4Y9SGiW4IUb8G1SmAvcJSIzgc7AERtPMKGoKB/05/qhHhIfzKUlr1ZBTkIYPq9Ypzycdpiq5asS\nHhbOPe3voW6lurSu1ZqTGSdJmJPAnhN7qFmhJmO7ji38ZMXktaQgIjOAHkAtEUkFRgNlAVT1FWA+\n0BdIBk4CZ6dLYwKYpx/2Rfmgtw91P7HmTfhslONn9+6fum0d3/6LSFX5bNtnjF89nlEdRjG4xWB6\nNurper1i2Yq83PNlHlj6AM/+5Vkqlq14rneQL2/OPrqmkNcVuNNb1zfGG7zxrd4+6P1cXi2CHc7x\ngL+9kGf3T1HsObGHccvH8e2ub7nwvAtpX7t9nsc1q96MTwZ8ck7X8kTAlc42xls8+cC3b/VBqLDB\n4R15DAgXMB5QFPO3zWfcinFkazYPdnyQa1peQ3hY+Dmd81xZUjBBryS7ceyDPki4J4K8PvTdlVAC\nyEuV8lVoW6sto7uMJioyqsTPXxzi6MUJHPHx8Wqb7JjcCvrgL8q3e/vAD0IFdf/kJAIvfejnlpmd\nyTtJ75CRncGIC0cAjvEEkbxm6JcsEVmrqvGFHWctBRPQcpJBQR/89u0+xOROAl7s/imKrQe3krgs\nkaQDSVwRc4UrGZRGQigKSwomIOTXEnBPBvbBH6SKsiAMzk4CPkgA7tKz0nl146tM3TSVKuWr8Fz3\n5+jdqLffJYMclhSM35u+cif/nr0JOLslYMkggHn6YV9Yn39uPk4Cue04uoOpm6fSt0lf/i/+/6gW\nUc3XIRXIkoLxC56MCTw1sK19+AeL/Ob558XPPuQ9cTLjJF+mfMnfmvyN5tWbMzdhLg0jGxb+Rj9g\nScGUurwSgI0JBDEvz/P3N8t+X8a45eP4/fjvtKrRiibVmgRMQgBLCqYUFTQobB/8QcRPB3q97cjp\nIzy35jlmJ88mpkoMb/Z5kybVmvg6rCKzpGBKRe5xAUsAfqioA7r58bOB3tKQlZ3FDQtuYMfRHdza\n9lZGthtJ+fDyvg6rWCwpGK/I3UVk4wJ+qiiLuDwVAkkgx6G0Q38WsOtwD/Uq1aNVzVa+DuucWFIw\nJSK/JJDTRWStAz+TkwzcE0EIfZifK1Xl022fMn7VeEbFjeLqFlfTM7pn4W8MAJYUzDnJb5zAkoCf\ny6n7b4mgyH4//jvjlo/j+9+/J/a8WOLqxPk6pBJlScEUmXurwBaPBQAv1P0PVZ/++ilPrHgCRXm4\n08MMazmMMAnzdVglypKCKVRBXUOWDPxUYWMFxaz7H+pqRNSgfe32JHZJpH7l+r4OxyusIJ7JV0FT\nSC0R+KGCEoF1ERVLRnYGb215i8zsTEa2GwmUXgG7kmYF8cw5sSmkAaCg9QA2VnDOfjzwI6OXjebH\ngz9yZcyVflvArqRZUjBnyN06sCmkfiz3JvGWCErE6azTvLLhFd7c/CbVyldjYo+J9GrUy9dhlRpL\nCsbFWgcBwL11YIPFXrHz6E6mbZlGv6b9eCD+AaqWr+rrkEqVJQVjrQN/lN/qYvcuIhssLjEnM07y\nxc4v6Ne0H82rN+fThE/9Zie00mZJIQQVNJvIWgc+4MnOYDmsi6jEfb/re8YuH8ueE3toXbM1Tao1\nCdmEAJYUQkZ+awty/mvJwIdyjw2AffiXgsNph5mwZgJzf51L46qNeevKtwKygF1Js6QQ5PKaVmpJ\nwA/Z2ECpysrO4voF15NyLIXb2t7GP9r9I2AL2JU0SwpBKq9kYInARwqrPpq7lWC85mDaQaqVr0Z4\nWDj3xd1H/cr1aVmjpa/D8iuWFIKMJQM/UZTqozZg7HWqyifJnzBhzQRGdRjFkPOHcFn0Zb4Oyy9Z\nUggSlgz8SO6tJm18wKd2Hd/F2GVjWb57OR1qd6BT3U6+DsmvWVIIAra+wI+4J4Qg3Goy0Hz666c8\nvuJxBOHRzo9y9flXB10Bu5JmSSEI5MwqsvUFPpBfqQlLCH6hZkRN4urEkXhRIvUq1/N1OAHBkkIA\nyr3OIGn3UTo3rmEJwds8WU9gXUU+lZGdwZub3yRLs7i93e10bdCVrg26+jqsgGJJIYDkV7W0Vb0q\nDIht4MvQQoOtJ/BrSQeSSPw+ka2HttK3cd+ArWbqa5YUAoANIvuQ1Rrye2mZafx3w395a8tbVI+o\nzguXvhA0W2P6gleTgoj0AV4EwoHXVfXpXK9HA28B1ZzHPKSq870ZU6CxQWQfyWsPY5s66pdSj6Xy\ndtLbDGg2gPvj7g+5AnYlzWtJQUTCgSlAbyAVWC0ic1U1ye2wR4H3VfW/ItIKmA/EeCumQOOeEGwQ\n2YsKGyuw7iG/czz9OEt2LiGhWQLNqjdj3sB5QbsTWmnzZkuhE5CsqtsARGQmMABwTwoKVHH+XBX4\n3YvxBBybVeQlBW1Ok8OSgd9amrqUx1c8zt6Te7mw1oU0qdbEEkIJ8mZSaACkuD1OBTrnOmYM8LmI\n3A1UAvLcyUJERgAjAKKjQ+PDcfrKnaz87aDNKippuReW5fzXEoDfO5R2iGdWP8Nn2z6jadWmvH3l\n21bAzgu8mRTyGvbPvSH0NcA0VX1ORLoA74hIG1XNPuNNqq8Br4Fjj2avROsncg8q26yiEmQLywJW\nVnYWNyy4gdRjqYxsN5Lb2t5GufByvg4rKHkzKaQCDd0eR3F299AtQB8AVV0uIhFALWCvF+PyWzao\nXMJsYVnA239qPzUiahAeFs4/4/9JvUr1OL/G+b4OK6h5MymsBpqLSGNgFzAMuDbXMTuBnsA0EbkA\niAD2eTEmv2WDyiWkoEJ01k0UMFSV2cmzeXb1s4yKcxSw69Gwh6/DCgleSwqqmikidwGLcEw3naqq\nW0RkHLBGVecC/wT+JyL34ehauklVg7p7KC+WEM5RfonAkkBASjmWwthlY1m5ZyXxdeK5qN5Fvg4p\npEigfQbHx8frmjVrfB1GibGEUAyFzR6yRBCw5iTP4cmVTxImYdwfdz+DWwy2AnYlRETWqmp8YcfZ\nimYfsoRQDDZ7KKidV/E8OtXtxKMXPUrdSnV9HU5IsqTgI5YQPGSDxUEtIyuD1ze/jqpyR+wddK3f\nla71rYCdL1lSKGW5p5xaQiiAtQqC2ub9m3ns+8dIPpxMvyb9rICdn7CkUMrmrN/lKnVtU04LkdNC\nsFZBUDmVeYop66bwzo/vUKtCLV667CWbWeRHLCmUIvdVyrP+0cXX4fin3FVJG3WzhBBkdh3bxfSf\npjOo+SDui7uPyHKRvg7JuLGkUIpyahnZKuU8WFXSoHYs/RhLdixhYPOBNKvejPlXzbeBZD9lSaGU\nWC2jAuQeO7Axg6CyNHUpY5ePZf+p/bSr3Y4mVZtYQvBjlhRKgftMo5BvJRRUptrGDoLKwbSDjF81\nnvm/zadZtWa80OMFmlS1Anb+zpKCl9nUUzd5zSbK+dlaB0ElKzuLGxfcSOrxVO6IvYNb29xK2fCy\nvg7LeMCjpCAi5YBoVU32cjxBxRJCLjabKOi5F7B7IP4B6leuT/PqzX0dlimCQtePi8hfgU3AYufj\nWBGZ7e3AgoFtkuO05k148682myiIZWs27299n7/N/hsfbP0AgO4Nu1tCCECetBTG4dgc5ysAVV0v\nIs28GlWAy1mglrMeIeQSQkG1iWw2UdDZeXQnY5aPYfWe1XSu25muDWxFciDzJClkqOrhXCsNA6uK\nXinKa0+EkGKrkEPK7F9m8+TKJykbVpYxXcZwVfOrbFVygPMkKfwoIkOAMOfeCPcCK7wbVmAK+TEE\n29ks5NSrXI+u9bvySOdHqFOpjq/DMSXAk6RwF5AIZAMf49gf4WFvBhVorJ4RlhBCRHpWOq9vep1s\nzeau9ndxUb2LbL+DIONJUrhCVR8EHsx5QkSuwpEgQl5Ib6GZ1+Y2lhCC1sZ9Gxm9bDTJh5Pp37S/\nFbALUp4khUc5OwE8ksdzISkkZxjlVZLCxg2C1smMk0xeP5l3k96ldsXaTOk5hb9E/cXXYRkvyTcp\niMgVQB+ggYg87/ZSFRxdSSEv5EpX5JUMLBEEvd0ndjPrp1kMOX8IozqMonK5yr4OyXhRQS2FvcBm\nIA3Y4vb8MeAhbwYVCEKudIXVJwopR9OPsnj7Yga1GETTak2Zd9U8q1cUIvJNCqq6DlgnIu+palop\nxuT3QnKWka1GDhlf7vySJ1Y8wcG0g7Sv094K2IUYT8YUGojIk0ArICLnSVVt4bWo/FjIJYScLiNb\njRz0Dpw6wNOrnmbh9oW0qN6Cly57yQrYhSBPksI04AngWeBKYDghPKYQEgPLec0qstXIQS0rO4sb\nFtzA7hO7ubv93QxvM5yyYVbALhR5khQqquoiEXlWVX8FHhWRb70dmD8KiYHl3GMHNn4Q1Pae3Eut\nCrUIDwvnwU4P0qByA5pWa+rrsIwPeZIUTotjMvKvIjIS2AXU9m5Y/idoB5bzq1NkYwdBLVuz+WDr\nB0z8YSKjOoxiWMthNs3UAJ4lhfuAysA9wJNAVeBmbwblj4Ky28jqFIWk7Ue2M2b5GNb+sZaL6l1E\ntwbdCn+TCRmFJgVVXen88RhwPYCIRHkzKH8VNN1GudcbWKsgZHz8y8c8tfIpyoWXY1zXcSQ0S7BV\nyeYMBe6nICIdRSRBRGo5H7cWkbcJsYJ4OWMJQcN9NpElhJBSv3J9ujXoxpwBcxjYfKAlBHOWglY0\n/wcYBGzAMbg8G0eF1PHAyNIJzz/kdB0F/FiC+/TSum1h+DxfR2S8LD0rnVc2vALAPR3usQJ2plAF\ndR8NANqp6ikRqQH87ny8tXRC8y9B0XXknhBsemnQW793PYnLEvntyG8MbDbQCtgZjxSUFNJU9RSA\nqh4UkZ9CMSG4T0MNWNZCCCknM04yad0kpv84nbqV6vJKr1e4uMHFvg7LBIiCkkITEcmphCpAjNtj\nVPWqwk4uIn2AF4Fw4HVVfTqPY4YAY3Ds5rZBVa/1PHzvCpppqNZCCCm7T+zmg60fMKzlMO7tcC+V\nylbydUgmgBSUFAblejy5KCcWkXBgCtAbSAVWi8hcVU1yO6Y5jg17LlbVQyLiV+sfgmoaqrUQgtqR\n00f4fMfnXN3iappWa8qCQQuoXdGv/ncyAaKggnhfnOO5OwHJqroNQERm4hinSHI75jZgiqoecl5z\n7zles8QF9FhC7m4jE5S+2PEFT6x8gkNph4ivE0/jqo0tIZhi82TxWnE1AFLcHqcCnXMd0wJARL7H\n0cU0RlUX5j6RiIwARgBERwfoB3Rpym/fAxNU9p/az1Mrn2LxjsW0rNGSKT2n0LhqY1+HZQKcN5NC\nXtMcNI/rNwd6AFHAtyLSRlUPn/Em1deA1wDi4+Nzn8MrAnaA2fY9CAlZ2VncuOBG9pzYw70d7uXG\n1jdaATtTIjxOCiJSXlVPF+HcqUBDt8dROKa15j5mhapmAL+JyFYcSWJ1Ea7jFQG7NsH2PQhqe07s\noXbF2oSHhfNQp4doENnAylubElXgimYAEekkIpuAX5yP24nISx6cezXQXEQai0g5YBgwN9cxnwCX\nOs9bC0d30rYixO9VATueYPseBJ1szea9H9+j/yf9mbV1FgCXRF1iCcGUOE9aCpOAv+H4AEdVN4jI\npYW9SVUzReQuYBGO8YKpqrpFRMYBa1R1rvO1y0UkCcgC/k9VDxTzXowJStuObGPMsjGs27uOi+tf\nTPeo7r4OyQQxT5JCmKruyLUSMsuTk6vqfGB+rucS3X5W4H7nH3Ou1rzpGFxuZFUvg8VHP3/EUyuf\nIqJMBE92e5J+TfrZqmTjVZ4khRQR6QSoc+3B3cDP3g3LtwJqkDmvXdJsplHQaBjZkO4Nu/Pvzv+m\nVoVavg7HhABPksLtOLqQooE/gCXO54JWwAwy2y5pQed01mlXAbt7O9xLp3qd6FSvk4+jMqHEk6SQ\nqarDvB6JnwmIQWabaRRU1u1dR+L3iWw/up1BzQdZATvjE54khdXOqaKzgI9V9ZiXY/KpgOk6ch8/\nsIQQ0E5knODFH15k5k8zqV+5Pq/2epWuDbr6OiwTogqdkqqqTYEngDhgk4h8IiJB23IImK6jnFaC\njR8EvD9O/MHHv3zMtRdcy8f9P7aEYHzKo8VrqroMWCYiY4AXgPeAmV6My6f8tuvIfVA5Z+c0ayUE\npMNph1m0fRFDWw6lSbUmLLhqAedVPM/XYRlTeFIQkco4CtkNAy4A5gD2Vaa05R5UtjLYAUlVWbxj\nMU+ufJKjp4/SqV4nGldtbAnB+A1PWgqbgU+BZ1T1Wy/H41N+PZ5gg8oBb9/JfTy58km+2PkFrWq2\n4rXer1kBO+N3PEkKTVQ12+uR+AG/H0+w7qKAlZWdxY0Lb2Tvyb3cH3c/17e6njJh3qxHaUzx5Puv\nUkSeU9V/Ah+JyFmVST3ZeS2QuLcS/G48wVYqByz3AnaPdH6EBpUbEFM1xtdhGZOvgr6qzHL+t0g7\nrgUqv20luI8l2BhCwMjKzmLm1pm8+MOL3Bd3H9e0vMb2STYBoaCd11Y5f7xAVc9IDM5Cd+e6M5vf\n8atWQu6NcmwsIWBsO7yNxGWJbNi3gW4NutEjqoevQzLGY4WuUwBuzuO5W0o6EF/K6TryGzmtg5wu\nI0sIAeODnz9g8KeD2XF0B091e4qXe75Mvcr1fB2WMR4raExhKI5pqI1F5GO3lyKBw3m/KzD5VdeR\ne3eRJYOA0yiyET2je/JQp4eoWaGmr8MxpsgKGlNYBRzAsWPaFLfnjwHrvBlUafK7AWabehpQ0jLT\neHnDywjCfXH3WQE7E/AKGlP4DfgNR1XUoOU3rYScMQRbqRww1uxZw5jlY9hxdAdDWgyxAnYmKBTU\nffSNqnYXkUOA+5RUwbE/jh+u8Coev2gl5CQEW6ns946nH+eFH15g1tZZRFWO4vXLX6dzvc6+DsuY\nElFQ91HOlptBu7OH36xgdl+HMHyeb2Mxhdp7ai9zkudwQ6sbuDP2TiqWrejrkIwpMQV1H+WsYm4I\n/K6q6SLSDbgQeBc4WgrxeZVfdB3ZOoSAcCjtEIu2L2JYy2E0qdqEBYMW2E5oJih5MiX1ExxbcTYF\n3sZRFG+6V6MqRT7tOrKZRn5PVVn420IS5iQwfvV4th/ZDmAJwQQtT4qvZKtqhohcBbygqpNEJGhm\nH/mUzTTya3tP7uXxFY/zdcrXtK7Zmv9d/j8rUWGCnkfbcYrI1cD1QILzubLeC6l0+Hw8wXZO82tZ\n2VnctPAm9p7cywPxD/D3C/5uBexMSPDkX/nNwB04SmdvE5HGwAzvhuV9Ph9PsJ3T/NLvx3+nTsU6\nhIeF82jnR4mKjCK6ih+sXzGmlHiyHedm4B5gjYi0BFJU9UmvR1YKfD4V1VoJfiMrO4u3trzFgE8G\nMGuroxZk1wZdLSGYkOPJzmuXAO8Au3CsUagrIter6vfeDs6Y0vDLoV8YvWw0m/ZvontUdy6LvszX\nIRnjM550H00E+qpqEoCIXIAjScR7MzBv8vl4gvEb7299n/+s+g+RZSMZf8l4rmx8pa1KNiHNk6RQ\nLichAKjqjyJSzosxeZ1PxhNyyljkyFm9bHwipyRF46qNubzR5TzY6UFqRNiXBGM8SQo/iMirOFoH\nAH8nCArilep4gvt6hJzd06ychU+cyjzFlHVTCAsL4/64++lYtyMd63b0dVjG+A1PksJIHAPN/8Ix\nprAUeMmbQQUdW4/gF1bvWc3oZaNJOZbC0POHWgE7Y/JQYFIQkbZAU2C2qj5TOiEFKZtp5DPH0o/x\n/Nrn+fDnD2kY2ZA3Ln/Dylsbk498p6SKyL9xlLj4O7BYRPLagc0UJmeRmvGZfaf2MW/bPG5qfRMf\n9f/IEoIxBShoncLfgQtV9WqgI3B7UU8uIn1EZKuIJIvIQwUcN1hEVEQCdkZTvmyRmk8cTDvIez++\nB0CTqk1YOGgh/4z/JxXKVPBxZMb4t4K6j06r6gkAVd0nIp4Uz3MRkXAcO7b1BlKB1SIy130mk/O4\nSBxjFiuLFHkgsFIWpU5Vmf/bfJ5e9TTHM45zcf2LiakaYzOLjPFQQUmhidvezAI0dd+rWVWvKuTc\nnYBkVd0GICIzgQFAUq7jHgeeAR4oSuDFVaprFKyVUKr2nNjD4yseZ2nqUi6sdSFju461AnbGFFFB\nSWFQrseTi3juBkCK2+NU4IztqUSkPdBQVT8TkXyTgoiMAEYAREef2zTSUlujYK2EUpWZncnwhcM5\nkHaAf3X8F9e2vJbwsHBfh2VMwClok50vzvHcec31c23r6eyOmgjcVNiJVPU14DWA+Ph4LeTwQpXK\nGgVrJZSKXcd3UbdiXcqElSGxSyJRkVE0jGzo67CMCVhFGicoolQcu7bliAJ+d3scCbQBvhaR7cBF\nwNygGmy2VoLXZGZnMm3zNAZ8MoCZW2cC0KV+F0sIxpwjbxaIXw00d5ba3gUMA67NeVFVj+C2/7OI\nfA08oKprvBiTd7mXsrAyFl6z9eBWRi8bzZYDW7i04aX0btTb1yEZEzQ8TgoiUl5VT3t6vKpmishd\nwCIgHJiqqltEZBywRlXnFj1cP5a7lIWVsfCKmT/NZPyq8VQpX4UJ3SdwRaMrbFWyMSXIk9LZnYA3\ngKpAtIi0A25V1bsLe6+qzgfm53ouMZ9je3gSsN+yUhZelVOSolm1ZvRp3Id/dfwX1SOq+zosY4KO\nJy2FScDfcKxuRlU3iMilXo0qUNkYQok7mXGSl9a9RJmwMvwz/p/E140nvm7wDDsZ4288GWgOU9Ud\nuZ7L8kYwxrhbsXsFV829ind/fJf0rHRUz3nimTGmEJ60FFKcXUjqXKV8N/Czd8MKMO5rEsw5O5p+\nlOfWPMfHv3xMoyqNmNZnGnF14nwdljEhwZOkcDuOLqRo4A9gCcWogxTUbE1CiTpw6gALflvAzW1u\n5vZ2txNRJsLXIRkTMgpNCqq6F8d0UlMQG084J/tP7Wfhbwu5rtV1NK7amEWDFtlAsjE+4Mnso//h\nthI5h6qO8EpEJqSoKp9t+4zxq8dzMuMkl0RdQqMqjSwhGOMjnnQfLXH7OQIYyJk1jYwplt3HdzNu\nxTi+2/Ud7c5rx7iu42hUpZGjQf7lAAAe0klEQVSvwzImpHnSfTTL/bGIvAMs9lpEJiRkZmcyfNFw\nDqYd5KFODzHs/GFWwM4YP1CcMheNAfs6l8NmHhVJyrEU6leqT5mwMozpOoaGkQ1pUNnLFWuNMR4r\ndJ2CiBwSkYPOP4dxtBL+7f3QAoTNPPJIZnYmb2x6g4RPElwF7C6qd5ElBGP8TIEtBXEUlWmHo6Ad\nQLYG8Aoir22wYzOPCvTTwZ9I/D6RHw/+SM/onlze6HJfh2SMyUeBSUFVVURmq2pQrBwqtQ12jMv0\nH6czYfUEqpavyvM9nreKpsb4OU/KXKwSkQ5ej6SUlOgGOznjCeYsOQ3KFtVb0LdJX+YkzLGEYEwA\nyLelICJlVDUT6AbcJiK/Aidw7Kimqho0iaJY3Etl23iCy8mMk0xaN4kyUoYHOj5gBeyMCTAFdR+t\nAjoACaUUS+BwTwhWKttl2a5ljF0+lt0ndnPtBde6yl0bYwJHQUlBAFT111KKJXDY3glnOHL6CBNW\nT2DOr3OIqRLDtD7T6FAntBuSxgSqgpLCeSJyf34vqurzXogncNiMI5eDaQdZvGMxt7a9lZHtRlI+\nvLyvQzLGFFNBSSEcqIyzxWCMu/2n9jN/23xuaH2Dq4BdtYhqvg7LGHOOCkoKu1V1XKlFEihCfAWz\nqjL317k8s/oZ0jLT6N6wO42qNLKEYEyQKHRMwbgJ8RlHu47vYtzycSz7fRnta7dnTNcxVsDOmCBT\nUFLoWWpRBIIQn3GUmZ3JLYtu4VDaIR7p/AhDzh9CmHiyzMUYE0jyTQqqerA0A/Fba950zDbKWaQW\nYglh59GdNKjcgDJhZRjXdRxRkVHUr1zf12EZY7zEvuoVZtOHsGeTYwwhhBJCRnYG/9v4PxLm/FnA\nrlO9TpYQjAlyxSmdHRpyWgh7NkHdtjB8nq8jKjVJB5IYvWw0Px38icsbXc4VMVf4OiRjTCmxpJAf\n94QQQoPK7/34HhNWT6B6RHVe6PECPRvZ0JIxocSSQkFCqIWQU5KiZY2W9GvajwfiH6Bq+aq+DssY\nU8osKeQlhNYinMg4wQtrX6BceDn+r+P/EVcnjrg6QVEp3RhTDDbQnJcQ2U3tu13fMXDOQGZtnYWi\nBPD+ScaYEmIthfwEcW2jw2mHmbBmAnN/nUuTqk14+8q3ia0d6+uwjDF+wJJCCDp8+jBf7PyCf1z4\nD0ZcOIJy4eV8HZIxxk94tftIRPqIyFYRSRaRh/J4/X4RSRKRjSLyhYhYzQQv2XdyH9M2T0NViaka\nw6JBi7ir/V2WEIwxZ/BaUhCRcGAKcCXQCrhGRFrlOmwdEK+qFwIfAs94Kx6PBdkWm6rK7F9mM+CT\nAUxeP5mdx3YC2MwiY0yevNlS6AQkq+o2VU0HZgID3A9Q1a9U9aTz4QogyovxeCaIBplTj6UyYvEI\nEpcl0qJGCz7s96EVsDPGFMibYwoNgBS3x6lA5wKOvwVYkNcLIjICGAEQHR1dUvHlLwgGmTOzM7n1\n81s5fPowj130GINbDLYCdsaYQnkzKeRVejvPOY8ich0QD3TP63VVfQ14DSA+Pt478yZzl7UIUDuO\n7iCqchRlwsrw+MWP0zCyIXUr1fV1WMaYAOHNr46pQEO3x1HA77kPEpFewCNAf1U97cV4ChbgZS0y\nsjN4dcOrDJwzkBk/zQCgY92OlhCMMUXizZbCaqC5iDQGdgHDgGvdDxCR9sCrQB9V3evFWDwToGUt\ntuzfQuKyRH4+9DNXxlzJlY2v9HVIxpgA5bWkoKqZInIXsAjHfs9TVXWLiIwD1qjqXGACjn2gPxAR\ngJ2q2t9bMQWjd5PeZcKaCdSKqMWkSydxafSlvg7JGBPAvLp4TVXnA/NzPZfo9nMvb17fYwFY6yin\ngF3rWq0Z2Gwg98ffT5VyVXwdljEmwNmKZgioaajH048zce1EyoWX48FOD9K+dnva127v67CMMUHC\n5ijmCIBpqEtTl5IwJ4EPf/mQMmFlrICdMabEWUshABxKO8T41eOZt20ezao14/kez3PheRf6Oixj\nTBCypBAAjqYf5ZuUb7i93e3c1vY2yoaX9XVIxpggZUnBT/1x4g/m/TaP4a2H06hKIxYNXmQDycYY\nr7Ok4GdUlY9++Yjn1jxHZnYmvaJ7EV0l2hKCMaZUWFLwIylHUxizfAyr9qyiY92OjOkyhugqpVDr\nyQSFjIwMUlNTSUtL83UoxociIiKIioqibNnidTNbUvATOQXsjqQfIbFLIoOaD7ICdqZIUlNTiYyM\nJCYmBudiUBNiVJUDBw6QmppK48aNi3UOSwo+Xrj225HfaBjZkDJhZXii2xNWwM4UW1pamiWEECci\n1KxZk3379hX7HPZV1EcL1zKyMvjv+v9y1dyrmPnTTMAK2JlzZwnBnOu/gdBuKbi3Ekpx4dqmfZtI\nXJZI8uFk+jbuy1+b/LXUrm2MMQUJ7ZaCD1oJ7yS9w3ULruNo+lEmXzaZ8X8ZT/WI6qV2fWO85cCB\nA8TGxhIbG0vdunVp0KCB63F6erpH5xg+fDhbt24tsZjS09OpUaMGjz322BnPR0VFcfjwYdfjJUuW\nkJCQ4Ho8b9484uLiaNWqFS1btuTBBx/M8/xPPPEEzZo1o2XLlixZsiTPY5YsWUKHDh1o06YNN998\nM5mZmQBs2bKFLl26UL58eV544QXX8Tt27KBHjx60atWK1q1bM3nyZNdrP/zwA507dyY2NpaOHTuy\nZs2aov9SCqOqAfUnLi5Oi2vIK8t0yCvL/nxial/Hn1KQnZ2tqqrr/linY5eN1aOnj5bKdU3oSEpK\n8nUILqNHj9YJEyac9Xx2drZmZWWVWhxz5szRiy++WJs3b37G8w0aNNBDhw65Hi9evFgHDBigqqrr\n16/XZs2a6datW1VVNSMjQ19++eWzzr1hwwZt3769nj59WpOTk7VZs2Zn3VtmZqY2aNBAk5OTVVX1\n4Ycf1mnTpqmq6p49e3T16tX64IMP6sSJE13v2bVrl65bt05VVY8cOaJNmjRxxXLppZfq559/7rq3\nnj175nnfef1bwFGdutDP2NDuPioFx9KP8fza54kIj+DBTg8SWzuW2Nqxvg7LBLmxn24h6fejJXrO\nVvWrMLpf6yK/Lzk5mYSEBLp168bKlSv57LPPGDt2LD/88AOnTp1i6NChJCY6iid369aNyZMn06ZN\nG2rVqsXIkSNZsGABFStWZM6cOdSuXbtI154xYwb3338/EydOZPXq1XTs2LHQ94wfP57HHnuMFi1a\nAFCmTBluv/32s46bM2cO11xzDeXKlaNp06ZER0ezdu3aM66xd+9eKleuTNOmTQHo3bs3EydO5MYb\nb6ROnTrUqVOHTz755Izz1q9fn/r16wNQpUoVWrZsya5du2jRogUiwtGjjr/XI0eOuI4rSaHdfeRl\nX6d8TcInCXz8y8eUDS9rBexMyEpKSuKWW25h3bp1NGjQgKeffpo1a9awYcMGFi9eTFJS0lnvOXLk\nCN27d2fDhg106dKFqVOnFumaJ06c4JtvvqFv375cc801zJgxw6P3bd68mbi4uDxfmz17NuPGjQNg\n165dNGz45+aSUVFR7Nq164zj69Spw8mTJ1m3bp1jYepHH5GSkoKntm3bxubNm12JZtKkSdx77700\nbNiQhx9+mCeffNLjc3kqZFoK01fuZOVvB+ncuIbXr3Uw7SBPr3qaBb8toHn15rx42Yu0qdXG69c1\nJkdxvtF7U9OmTc/4Bj1jxgzeeOMNMjMz+f3330lKSqJVq1ZnvKdChQpceaVjF8G4uDi+/fbbIl1z\n7ty59O7dm4iICK6++mri4+N59tlnCQsLy3OGjiezdgYOHMjAgQMB8vySl/scYWFhTJ8+nbvvvpv0\n9HR69epFmTKefewePXqUQYMG8dJLL1G5cmUApkyZwpQpUxgwYADTp0/ntttuY+HChR6dz1Mh01KY\ns96RwQfENvD6tY6nH+e71O+4I/YOZv11liUEE/IqVark+vmXX37hxRdf5Msvv2Tjxo306dMnz1XY\n5cqVc/0cHh7uGqB116tXL2JjYxk5cuRZr82YMYOFCxcSExNDx44d2bt3L0uXLgWgZs2aHDp0yHXs\nwYMHqVWrFgCtW7dm7dq1hd5TVFTUGd/6U1NT8+zO6datG9999x2rVq2iW7duNG/evNBzp6enc9VV\nV3HTTTfRv/+fm1G+++67DBgwAIChQ4eyYsWKQs9VVCGTFAA6N67BtZ29UzZiz4k9vL7pdVSV6CrR\nLBq8iNvb3W4VTY3J5ejRo0RGRlKlShV2797NokWLin2uJUuWsH79el555ZUznj906BArV64kNTWV\n7du3s337diZNmuTqQurRowfvvPMOAJmZmbz33ntceqljK9t//etfPPHEEyQnJwOQlZXF888/f9a1\n+/fvz4wZM0hPT+fXX39lx44deXY77d3r2H4+LS2NZ555Js8E5k5Vuemmm4iNjeXee+8947U6derw\n3Xffue79/PPPL/R3VFQh033kLdmazYc/f8jza58nW7O5vNHlRFeJJrJcpK9DM8YvdejQgVatWtGm\nTRuaNGnCxRdfXOLX+Oijj+jdu/cZ9X8SEhJ45JFHmDx5MmPGjGHkyJG0a9cOVXWNOwC0b9+eZ599\nliFDhnDq1ClExPXtfPbs2WzatInExETatWtHQkICF1xwAWXKlOHll18mLMzxPfuKK67gnXfeoXbt\n2vznP/9h4cKFZGdnc9ddd/GXv/wFcLQsLrroIo4ePUpYWBjPPvssP//8M6tWrWLGjBlceOGFxMY6\nJqWMHz+eK664gjfeeIN77rmHrKwsKlSowKuvvlrivzsJtMHP+Ph4Lc7c3KGvLgdg1j+6/Pnkm85F\nY8PnFSuWHUd3MGbZGNb8sYbO9TozustoGkY2LPyNxnjBjz/+yAUXXODrMIwfyOvfgoisVdX4wt5r\nLYViyszOZMTnIziWfoxxXceR0CzBSgwYYwJe6CaFYhbC23Z4G9FVoikTVoanLnmKhpENqV2xaHOn\njTHGX4XUQPMZiljiIj0rnSnrpzBo7iBm/OQYrIqrE2cJwRgTVEK3pQAeF8LbsG8Do78fza9HfqVf\nk370a9KvFIIzxpjSF9pJwQNvbXmL59Y8R51KdXi558tcEnWJr0MyxhivsaSQj2zNJkzCaHdeO4ac\nP4RRHUZRuVxlX4dljDFeFbpjCvk4mn6UxO8TeXrV0wDE1o7l0YsetYRgTCFKonQ2wNSpU9mzZ0+x\nYli9ejUiwhdffOF6Ljk52TXfP8ejjz7qKletqjzzzDOcf/75tGnThtjYWN57772zzp2WlsbgwYNp\n1qwZXbp0YefOnWcdk5SU5Lrn2NhYIiMjzyh9PXHiRM4//3xatWrFv//9b9fz+ZXgjoqKom3btsTG\nxtK5c+di/U6KKjRbCvnMPPpi5xc8ueJJDqYdZHib4aiqTTM1xkM1a9Zk/fr1AIwZM4bKlSvzwAMP\nFPk8U6dOpUOHDtStW/RdCGfMmEG3bt2YMWMGPXv29Og9U6ZM4auvvmLNmjVERkZy+PBh5s6de9Zx\nr732GnXr1iU5OZl3332Xhx9++Kzk0apVK9fvIDMzk/r167v2aVi8eDELFy5k48aNlC9f3rXSeePG\njXz88cckJSWRkpJCnz592Lp1q2sh3Lfffku1atWK/LsortBMCrlmHh04dYCnVj7F5zs+p2WNlkzu\nOZlWNVsVcAJj/NyCh2DPppI9Z922cOXTxXrrW2+9xZQpU0hPT6dr165MnjyZ7Oxshg8fzvr161FV\nRowYQZ06dVi/fj1Dhw6lQoUKrFq16owaSAXJzs7mo48+4quvvuKSSy4hPT3do/c+9dRTLF++nMhI\nRxWCatWqccMNN5x13Jw5c3j6acf9DxkyhFGjRhV43s8//5wLLriAqKgoAP773//y8MMPU758eQBX\nGXBPSnCXptDrPspjC84TGSdYvns597S/h+l/nW4JwZgStHnzZmbPns2yZctYv349mZmZzJw5k7Vr\n17J//342bdrE5s2bueGGGxg6dCixsbHMmjWL9evXe5wQAJYuXUrLli1dpTM8qR566NAhMjIyaNSo\nUZ6vP/LII8yfPx84s1R2uXLlqFSp0hm7t+U2c+ZMV+kMgJ9//pmvv/6azp0706NHD1fRvYJKcIsI\nl112GXFxcbzxxhuF3k9JCJmWQpr8zq6yr5G8KZtmwO7zL+fTja9xW9vbiK4SzeLBi6lUtlKh5zEm\nIBTzG703LFmyhNWrVxMf76iwcOrUKRo2bMgVV1zB1q1buffee+nbty+XX375OV1nxowZDBs2DIBh\nw4YxY8YM+vfvn28XsIgUuseJ+34FnpTKzpGWlsa8efPOKKSXmZnJkSNHWLlyJcuXL2fo0KEkJycX\neN6VK1dSv3599uzZQ+/evbngggvo2rVrgTGfK6+2FESkj4hsFZFkEXkoj9fLi8gs5+srRSTGG3Gc\nzDhJSrlJnJbfuSMrlbcbXUjCb9N5fdPrpBxzlL61hGCMd6gqN998M+vXr2f9+vVs3bqVxx57jJo1\na7Jx40a6devGpEmT+Mc//lHgebZv3+4awH399dfPeC0jI4PZs2eTmJhITEwMo0aNYt68eZw4ceKs\nMtnwZ6nsGjVqULZs2TwHjXNzL5Wdnp7OiRMnqFq1ap7Hzps3j86dO7vKcee8f9CgQQB06dKFjIwM\nDh06VGAJ7pz/1q1blwEDBrBq1apC4zxXXksKIhIOTAGuBFoB14hI7n6ZW4BDqtoMmAiM90YsicsS\nyeQYCOwpE86EsMO0O68dswfMJrqKd0ppG2McevXqxfvvv8/+/fsBxyylnTt3sm/fPlSVq6++2rU9\nJ0BkZCTHjh076zwxMTGuxHLrrbee8drnn39Ox44dSUlJYfv27ezcuZN+/foxd+5cqlWrRvXq1fnm\nm29c1//8889d1Vkfeugh7rjjDtc1Dx8+zP/+97+zrt+/f3/eeustAN5///0CWzYzZsw4o+sIHFVa\nv/zyS8BRsA6gevXq+ZbgPn78OMePHwfg+PHjLF68mDZtvL83ize7jzoByaq6DUBEZgIDAPd99wYA\nY5w/fwhMFhHREizdOvuX2SxNXYpKBgAqQtmwsvRp3IcGlb2/4Y4xoa5t27aMHj2aXr16kZ2dTdmy\nZXnllVcIDw/nlltucc3yGz/e8Z1w+PDh3HrrrUUaaJ4xY4ZrR7QcgwYN4s033+Saa67h3Xff5c47\n7+TIkSOAYwpoTEwMAHfffTcnTpwgLi6OcuXKUbZsWf71r38BjjGFiy++mL59+zJixAiuu+46mjVr\nRq1atZg5cyYAKSkp3Hnnna4ZS8ePH+err77izTffPCOe2267jZtuuok2bdpQvnx53n77bYB8S3Dv\n3r2bwYMdk2EyMzO5/vrr6dWrV3H+CorEa6WzRWQw0EdVb3U+vh7orKp3uR2z2XlMqvPxr85j9uc6\n1whgBEB0dHTcjh07PI6j+6zuHEw7eNbzNSJq8M3Qb4p8X8b4KyudbXKcS+lsb44p5DUCkzsDeXIM\nqvqaqsaravx5551XpCBGdRhFhTIVznguIjyC++LuK9J5jDEmFHgzKaQC7jvORAG/53eMiJQBqgJn\nf60/BwObD+QvUX+hfLhjbnD5sPL0aNiDhGYJJXkZY4wJCt5MCquB5iLSWETKAcOA3MsE5wI3On8e\nDHxZkuMJOcZ1HUeNiBoIQs0KNRnbdWxJX8IYvxBoOymakneu/wa8lhRUNRO4C1gE/Ai8r6pbRGSc\niPR3HvYGUFNEkoH7gbOmrZaEimUr8nLPl2lSrQlTek6hYtmK3riMMT4VERHBgQMHLDGEMFXlwIED\nREREFPscIbNHszHBLiMjg9TUVNLS0nwdivGhiIgIoqKiKFu27BnP2x7NxoSYsmXL0rhxY1+HYQJc\n6NU+MsYYky9LCsYYY1wsKRhjjHEJuIFmEdkHeL6k+Uy1gP2FHhVc7J5Dg91zaDiXe26kqoWu/g24\npHAuRGSNJ6PvwcTuOTTYPYeG0rhn6z4yxhjjYknBGGOMS6glhdd8HYAP2D2HBrvn0OD1ew6pMQVj\njDEFC7WWgjHGmAJYUjDGGOMSlElBRPqIyFYRSRaRsyqvikh5EZnlfH2liMSUfpQly4N7vl9EkkRk\no4h8ISKNfBFnSSrsnt2OGywiKiIBP33Rk3sWkSHOv+stIjK9tGMsaR78244Wka9EZJ3z33dfX8RZ\nUkRkqojsde5MmdfrIiKTnL+PjSLSoUQDUNWg+gOEA78CTYBywAagVa5j7gBecf48DJjl67hL4Z4v\nBSo6f749FO7ZeVwksBRYAcT7Ou5S+HtuDqwDqjsf1/Z13KVwz68Btzt/bgVs93Xc53jPfwE6AJvz\neb0vsADHzpUXAStL8vrB2FLoBCSr6jZVTQdmAgNyHTMAeMv584dATxHJa2vQQFHoPavqV6p60vlw\nBY6d8AKZJ3/PAI8DzwDBUE/ak3u+DZiiqocAVHVvKcdY0jy5ZwWqOH+uytk7PAYUVV1KwTtQDgDe\nVocVQDURqVdS1w/GpNAASHF7nOp8Ls9j1LEZ0BGgZqlE5x2e3LO7W3B80whkhd6ziLQHGqrqZ6UZ\nmBd58vfcAmghIt+LyAoR6VNq0XmHJ/c8BrhORFKB+cDdpROazxT1//ciCcb9FPL6xp973q0nxwQS\nj+9HRK4D4oHuXo3I+wq8ZxEJAyYCN5VWQKXAk7/nMji6kHrgaA1+KyJtVPWwl2PzFk/u+Rpgmqo+\nJyJdgHec95zt/fB8wqufX8HYUkgFGro9juLs5qTrGBEpg6PJWVBzzd95cs+ISC/gEaC/qp4updi8\npbB7jgTaAF+LyHYcfa9zA3yw2dN/23NUNUNVfwO24kgSgcqTe74FeB9AVZcDETgKxwUrj/5/L65g\nTAqrgeYi0lhEyuEYSJ6b65i5wI3OnwcDX6pzBCdAFXrPzq6UV3EkhEDvZ4ZC7llVj6hqLVWNUdUY\nHOMo/VU1kPdy9eTf9ic4JhUgIrVwdCdtK9UoS5Yn97wT6AkgIhfgSAr7SjXK0jUXuME5C+ki4Iiq\n7i6pkwdd95GqZorIXcAiHDMXpqrqFhEZB6xR1bnAGziamMk4WgjDfBfxufPwnicAlYEPnGPqO1W1\nv8+CPkce3nNQ8fCeFwGXi0gSkAX8n6oe8F3U58bDe/4n8D8RuQ9HN8pNgfwlT0Rm4Oj+q+UcJxkN\nlAVQ1VdwjJv0BZKBk8DwEr1+AP/ujDHGlLBg7D4yxhhTTJYUjDHGuFhSMMYY42JJwRhjjIslBWOM\nMS6WFIzfEZEsEVnv9iemgGNj8qsmWcRrfu2sxLnBWSLi/GKcY6SI3OD8+SYRqe/22usi0qqE41wt\nIrEevGeUiFQ812ub0GBJwfijU6oa6/Zneyld9++q2g5HscQJRX2zqr6iqm87H94E1Hd77VZVTSqR\nKP+M82U8i3MUYEnBeMSSggkIzhbBtyLyg/NP1zyOaS0iq5yti40i0tz5/HVuz78qIuGFXG4p0Mz5\n3p7OOv2bnHXuyzuff1r+3J/iWedzY0TkAREZjKO+1HvOa1ZwfsOPF5HbReQZt5hvEpGXihnnctwK\noYnIf0VkjTj2URjrfO4eHMnpKxH5yvnc5SKy3Pl7/EBEKhdyHRNCLCkYf1TBretotvO5vUBvVe0A\nDAUm5fG+kcCLqhqL40M51Vn2YChwsfP5LODvhVy/H7BJRCKAacBQVW2LowLA7SJSAxgItFbVC4En\n3N+sqh8Ca3B8o49V1VNuL38IXOX2eCgwq5hx9sFR1iLHI6oaD1wIdBeRC1V1Eo66OJeq6qXO0heP\nAr2cv8s1wP2FXMeEkKArc2GCwinnB6O7ssBkZx96Fo6aPrktBx4RkSjgY1X9RUR6AnHAamd5jwo4\nEkxe3hORU8B2HOWXzwd+U9Wfna+/BdwJTMaxP8PrIjIP8Lg0t6ruE5Ftzpo1vziv8b3zvEWJsxKO\nsg/uu24NEZEROP6/rodjw5mNud57kfP5753XKYfj92YMYEnBBI77gD+AdjhauGdtmqOq00VkJfBX\nYJGI3IqjzPBbqvqwB9f4u3vBPBHJc48NZz2eTjiKsA0D7gIuK8K9zAKGAD8Bs1VVxfEJ7XGcOHYg\nexqYAlwlIo2BB4COqnpIRKbhKAyXmwCLVfWaIsRrQoh1H5lAURXY7ayRfz2Ob8lnEJEmwDZnl8lc\nHN0oXwCDRaS285ga4vn+1D8BMSLSzPn4euAbZx98VVWdj2MQN68ZQMdwlO/Oy8dAAo59AGY5nytS\nnKqagaMb6CJn11MV4ARwRETqAFfmE8sK4OKcexKRiiKSV6vLhChLCiZQvAzcKCIrcHQdncjjmKHA\nZhFZD7TEsWVhEo4Pz89FZCOwGEfXSqFUNQ1HBcoPRGQTkA28guMD9jPn+b7B0YrJbRrwSs5Ac67z\nHgKSgEaqusr5XJHjdI5VPAc8oKobcOzNvAWYiqNLKsdrwAIR+UpV9+GYGTXDeZ0VOH5XxgBWJdUY\nY4wbaykYY4xxsaRgjDHGxZKCMcYYF0sKxhhjXCwpGGOMcbGkYIwxxsWSgjHGGJf/BzO5m2BvRI+m\nAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from mlxtend.classifier import StackingClassifier\n",
    "\n",
    "# 构建Stacking 模型\n",
    "s_clf = StackingClassifier(\n",
    "    classifiers=[\n",
    "        LogisticRegression(C = 0.05, penalty = 'l1'), \n",
    "        SVC(C=0.01, kernel='rbf', gamma=0.01, probability=True), \n",
    "        SVC(C=0.01, kernel='linear', probability=True), \n",
    "        DecisionTreeClassifier(class_weight='balanced', max_depth=5, min_samples_split=72,), \n",
    "        XGBClassifier()\n",
    "    ], \n",
    "    meta_classifier=LogisticRegression(), use_probas=True, verbose=3)\n",
    "\n",
    "s_clf.fit(X_train, y_train)\n",
    "\n",
    "# 预测\n",
    "y_train_pred = s_clf.predict(X_train)\n",
    "y_test_pred = s_clf.predict(X_test)\n",
    "# 预测概率\n",
    "y_train_proba = s_clf.predict_proba(X_train)[:, 1]\n",
    "y_test_proba = s_clf.predict_proba(X_test)[:, 1]\n",
    "# 模型评估\n",
    "model_metrics(X_train, y_train, X_test, y_test, y_train_pred, y_test_pred, y_train_proba, y_test_proba)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "特征选择前结果：\n",
    "Fitting 5 classifiers...\n",
    "Fitting classifier1: logisticregression (1/5)\n",
    "LogisticRegression(C=0.05, class_weight=None, dual=False, fit_intercept=True,\n",
    "          intercept_scaling=1, max_iter=100, multi_class='warn',\n",
    "          n_jobs=None, penalty='l1', random_state=None, solver='warn',\n",
    "          tol=0.0001, verbose=1, warm_start=False)\n",
    "[LibLinear]Fitting classifier2: svc (2/5)\n",
    "SVC(C=0.01, cache_size=200, class_weight=None, coef0=0.0,\n",
    "  decision_function_shape='ovr', degree=3, gamma=0.01, kernel='rbf',\n",
    "  max_iter=-1, probability=True, random_state=None, shrinking=True,\n",
    "  tol=0.001, verbose=1)\n",
    "[LibSVM]Fitting classifier3: svc (3/5)\n",
    "SVC(C=0.01, cache_size=200, class_weight=None, coef0=0.0,\n",
    "  decision_function_shape='ovr', degree=3, gamma='auto_deprecated',\n",
    "  kernel='linear', max_iter=-1, probability=True, random_state=None,\n",
    "  shrinking=True, tol=0.001, verbose=1)\n",
    "[LibSVM]Fitting classifier4: decisiontreeclassifier (4/5)\n",
    "DecisionTreeClassifier(class_weight='balanced', criterion='gini', max_depth=5,\n",
    "            max_features=None, max_leaf_nodes=None,\n",
    "            min_impurity_decrease=0.0, min_impurity_split=None,\n",
    "            min_samples_leaf=1, min_samples_split=72,\n",
    "            min_weight_fraction_leaf=0.0, presort=False, random_state=None,\n",
    "            splitter='best')\n",
    "Fitting classifier5: xgbclassifier (5/5)\n",
    "XGBClassifier(base_score=0.5, booster='gbtree', colsample_bylevel=1,\n",
    "       colsample_bytree=1, gamma=0, learning_rate=0.1, max_delta_step=0,\n",
    "       max_depth=3, min_child_weight=1, missing=None, n_estimators=100,\n",
    "       n_jobs=1, nthread=None, objective='binary:logistic', random_state=0,\n",
    "       reg_alpha=0, reg_lambda=1, scale_pos_weight=1, seed=None,\n",
    "       silent=True, subsample=1)\n",
    "[准确率]\n",
    "训练集：0.8801\n",
    "测试集：0.7758\n",
    "[精准率]\n",
    "训练集：0.8293\n",
    "测试集：0.5638\n",
    "[召回率]\n",
    "训练集：0.6575\n",
    "测试集：0.4693\n",
    "[f1 score]\n",
    "训练集：0.7335\n",
    "测试集：0.5122\n",
    "[auc]\n",
    "训练集：0.9297\n",
    "测试集：0.7679\n",
    "\n",
    "特征选择后结果：\n",
    "Fitting 5 classifiers...\n",
    "Fitting classifier1: logisticregression (1/5)\n",
    "LogisticRegression(C=0.05, class_weight=None, dual=False, fit_intercept=True,\n",
    "          intercept_scaling=1, max_iter=100, multi_class='warn',\n",
    "          n_jobs=None, penalty='l1', random_state=None, solver='warn',\n",
    "          tol=0.0001, verbose=1, warm_start=False)\n",
    "[LibLinear]Fitting classifier2: svc (2/5)\n",
    "SVC(C=0.01, cache_size=200, class_weight=None, coef0=0.0,\n",
    "  decision_function_shape='ovr', degree=3, gamma=0.01, kernel='rbf',\n",
    "  max_iter=-1, probability=True, random_state=None, shrinking=True,\n",
    "  tol=0.001, verbose=1)\n",
    "[LibSVM]Fitting classifier3: svc (3/5)\n",
    "SVC(C=0.01, cache_size=200, class_weight=None, coef0=0.0,\n",
    "  decision_function_shape='ovr', degree=3, gamma='auto_deprecated',\n",
    "  kernel='linear', max_iter=-1, probability=True, random_state=None,\n",
    "  shrinking=True, tol=0.001, verbose=1)\n",
    "[LibSVM]Fitting classifier4: decisiontreeclassifier (4/5)\n",
    "DecisionTreeClassifier(class_weight='balanced', criterion='gini', max_depth=5,\n",
    "            max_features=None, max_leaf_nodes=None,\n",
    "            min_impurity_decrease=0.0, min_impurity_split=None,\n",
    "            min_samples_leaf=1, min_samples_split=72,\n",
    "            min_weight_fraction_leaf=0.0, presort=False, random_state=None,\n",
    "            splitter='best')\n",
    "Fitting classifier5: xgbclassifier (5/5)\n",
    "XGBClassifier(base_score=0.5, booster='gbtree', colsample_bylevel=1,\n",
    "       colsample_bytree=1, gamma=0, learning_rate=0.1, max_delta_step=0,\n",
    "       max_depth=3, min_child_weight=1, missing=None, n_estimators=100,\n",
    "       n_jobs=1, nthread=None, objective='binary:logistic', random_state=0,\n",
    "       reg_alpha=0, reg_lambda=1, scale_pos_weight=1, seed=None,\n",
    "       silent=True, subsample=1)\n",
    "[准确率]\n",
    "训练集：0.8687\n",
    "测试集：0.7765\n",
    "[精准率]\n",
    "训练集：0.8139\n",
    "测试集：0.5661\n",
    "[召回率]\n",
    "训练集：0.6180\n",
    "测试集：0.4665\n",
    "[f1 score]\n",
    "训练集：0.7025\n",
    "测试集：0.5115\n",
    "[auc]\n",
    "训练集：0.9128\n",
    "测试集：0.7605"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.model_selection import StratifiedKFold\n",
    "    \n",
    "# dataset_blend_train = np.zeros((X.shape[0], len(clfs)))\n",
    "# dataset_blend_test = np.zeros((X_submission.shape[0], len(clfs)))\n",
    "\n",
    "\n",
    "#idx = np.random.permutation(y.size)\n",
    "#X = X[idx]\n",
    "#y = y[idx]\n",
    "print type(y_train)\n",
    "# print y_train\n",
    "print y_train[[665]]\n",
    "\n",
    "clfs[0].fit(X_train, y_train)\n",
    "        \n",
    "n_folds = 5\n",
    "skf = StratifiedKFold(5)\n",
    "    \n",
    "print \"Creating train and test sets for blending.\"\n",
    "    \n",
    "#for j, clf in enumerate(clfs):\n",
    "    #print j, clf\n",
    "# dataset_blend_test_j = np.zeros((X_submission.shape[0], len(skf)))\n",
    "for (train_index, test_index) in (skf.split(X_train, y_train)):\n",
    "        X_train_ = (X_train[train_index])\n",
    "        y_train_ = pd.Series((y_train[train_index]))\n",
    "        X_test_ = (X_train[test_index])\n",
    "        y_test_ = pd.Series((y_train[test_index]))\n",
    "\n",
    "        # print np.argwhere(np.isnan(X_train))\n",
    "        # print np.argwhere(np.isnan(X_train_))\n",
    "        \n",
    "        print type(y_train)\n",
    "        print y_train[[665]]\n",
    "        print type(y_train_)\n",
    "        print y_train_[[665]]\n",
    "        #print np.isfinite(y_train)\n",
    "        #print np.isfinite(y_train_)\n",
    "            \n",
    "        clfs[0].fit(X_train, y_train)\n",
    "        clfs[0].fit(X_train_, y_train_)\n",
    "        # y_submission = clf.predict_proba(X_test)[:, 1]\n",
    "        # dataset_blend_train[test, j] = y_submission\n",
    "        # dataset_blend_test_j[:, i] = clf.predict_proba(X_submission)[:, 1]\n",
    "# dataset_blend_test[:, j] = dataset_blend_test_j.mean("
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.13"
  },
  "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": {},
   "toc_section_display": true,
   "toc_window_display": true
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}