05.19_Pandas_高级.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 分类数据"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 背景和目的"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "示例1：使用`pd.unique()`和`pd.value_counts()`从数据中提取不同值，并分别计算频率"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-11-14T02:27:21.412457Z",
     "start_time": "2019-11-14T02:27:14.205316Z"
    }
   },
   "outputs": [],
   "source": [
    "import numpy as np; import pandas as pd"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-11-13T11:22:16.832624Z",
     "start_time": "2019-11-13T11:22:16.812537Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0     apple\n",
       "1    orange\n",
       "2     apple\n",
       "3     apple\n",
       "4     apple\n",
       "5    orange\n",
       "6     apple\n",
       "7     apple\n",
       "dtype: object"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "values = pd.Series(['apple', 'orange', 'apple',\n",
    "                        'apple'] * 2)\n",
    "values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-11-13T11:22:27.511426Z",
     "start_time": "2019-11-13T11:22:27.450979Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['apple', 'orange'], dtype=object)"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.unique(values)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-11-13T11:22:44.185017Z",
     "start_time": "2019-11-13T11:22:44.173474Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "apple     6\n",
       "orange    2\n",
       "dtype: int64"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.value_counts(values)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "许多数据系统（数据仓库、统计计算或其它应用）都发展出了特定的表征重复值的方法，以进行高效的存储和计算。在数据仓库中，最好的方法是使用所谓的包含不同值的维表(Dimension Table)，将主要的参数存储为引用维表整数键："
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "示例2：根据字段(values)从另一个表中读取数据(values)  \n",
    "**Note:s.take()**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-11-13T11:25:14.852725Z",
     "start_time": "2019-11-13T11:25:14.836507Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0    0\n",
       "1    1\n",
       "2    0\n",
       "3    0\n",
       "4    0\n",
       "5    1\n",
       "6    0\n",
       "7    0\n",
       "dtype: int64"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "0     apple\n",
       "1    orange\n",
       "dtype: object"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "values = pd.Series([0,1,0,0]*2)\n",
    "dim = pd.Series(['apple', 'orange'])\n",
    "display(values, dim)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-11-13T11:25:25.031874Z",
     "start_time": "2019-11-13T11:25:25.021563Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0     apple\n",
       "1    orange\n",
       "0     apple\n",
       "0     apple\n",
       "0     apple\n",
       "1    orange\n",
       "0     apple\n",
       "0     apple\n",
       "dtype: object"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dim.take(values)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "这种用整数表示的方法称为分类或字典编码表示法。不同值得数组称为分类、字典或数据级。本书中，我们使用分类的说法。表示分类的整数值称为分类编码或简单地称为编码。\n",
    "\n",
    "分类表示可以在进行分析时大大的提高性能。你也可以在保持编码不变的情况下，对分类进行转换。一些相对简单的转变例子包括：\n",
    "\n",
    "+    重命名分类。\n",
    "+    加入一个新的分类，不改变已经存在的分类的顺序或位置。\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Pandas的分类类型"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "pandas有一个特殊的分类类型(category)，用于保存使用整数分类表示法的数据"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-11-13T11:31:02.480402Z",
     "start_time": "2019-11-13T11:31:02.455433Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>basket_id</th>\n",
       "      <th>fruit</th>\n",
       "      <th>count</th>\n",
       "      <th>weight</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>apple</td>\n",
       "      <td>5</td>\n",
       "      <td>3.443358</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>orange</td>\n",
       "      <td>10</td>\n",
       "      <td>3.769315</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>apple</td>\n",
       "      <td>4</td>\n",
       "      <td>3.613849</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>apple</td>\n",
       "      <td>4</td>\n",
       "      <td>3.626218</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>apple</td>\n",
       "      <td>5</td>\n",
       "      <td>0.435954</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>5</td>\n",
       "      <td>orange</td>\n",
       "      <td>7</td>\n",
       "      <td>3.241405</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>6</td>\n",
       "      <td>apple</td>\n",
       "      <td>10</td>\n",
       "      <td>3.824108</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>7</td>\n",
       "      <td>apple</td>\n",
       "      <td>12</td>\n",
       "      <td>2.702095</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   basket_id   fruit  count    weight\n",
       "0          0   apple      5  3.443358\n",
       "1          1  orange     10  3.769315\n",
       "2          2   apple      4  3.613849\n",
       "3          3   apple      4  3.626218\n",
       "4          4   apple      5  0.435954\n",
       "5          5  orange      7  3.241405\n",
       "6          6   apple     10  3.824108\n",
       "7          7   apple     12  2.702095"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "fruits = ['apple', 'orange', 'apple', 'apple'] * 2 \n",
    "N = len(fruits)\n",
    "df = pd.DataFrame({\n",
    "    \"fruit\":fruits,\n",
    "    'basket_id':np.arange(N),\n",
    "    'count':np.random.randint(3,13,size=N),\n",
    "    'weight':np.random.uniform(0, 4, size=N)},\n",
    "    columns=['basket_id', 'fruit', 'count', 'weight'])\n",
    "df"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "示例1：将df['fruits']进行分类"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-11-13T11:32:16.158189Z",
     "start_time": "2019-11-13T11:32:16.146508Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0     apple\n",
       "1    orange\n",
       "2     apple\n",
       "3     apple\n",
       "4     apple\n",
       "5    orange\n",
       "6     apple\n",
       "7     apple\n",
       "Name: fruit, dtype: category\n",
       "Categories (2, object): [apple, orange]"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "fruit_cat = df['fruit'].astype('category')\n",
    "fruit_cat"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "得到的是一个pandas.Categorical实例"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-11-13T11:33:01.555490Z",
     "start_time": "2019-11-13T11:33:01.548030Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "pandas.core.arrays.categorical.Categorical"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "c = fruit_cat.values\n",
    "type(c)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "示例2：获取pd.Categorical的categories和code属性"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-11-13T11:33:58.121375Z",
     "start_time": "2019-11-13T11:33:58.113827Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['apple', 'orange'], dtype='object')"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "c.categories"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-11-13T11:34:02.208242Z",
     "start_time": "2019-11-13T11:34:02.200712Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0, 1, 0, 0, 0, 1, 0, 0], dtype=int8)"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "c.codes"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "示例3：将原先的df中fruit列转换为分类"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-11-13T11:35:07.959741Z",
     "start_time": "2019-11-13T11:35:07.948771Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0     apple\n",
       "1    orange\n",
       "2     apple\n",
       "3     apple\n",
       "4     apple\n",
       "5    orange\n",
       "6     apple\n",
       "7     apple\n",
       "Name: fruit, dtype: category\n",
       "Categories (2, object): [apple, orange]"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df['fruit'] = df['fruit'].astype('category')\n",
    "df.fruit"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "示例4：从python序列中直接创建pd.Categorical\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-11-13T11:35:57.698580Z",
     "start_time": "2019-11-13T11:35:57.688587Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[a, b, c, a, a, c]\n",
       "Categories (3, object): [a, b, c]"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.Categorical(['a','b','c','a','a','c'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "示例5：根据分类编码创建Categorical"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-11-13T11:37:47.041685Z",
     "start_time": "2019-11-13T11:37:47.030308Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[foo, bar, bar, baz, foo, foo, baz]\n",
       "Categories (3, object): [foo, bar, baz]"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "categorical = ['foo', 'bar', 'baz']\n",
    "codes= [0,1,1,2,0,0,2]\n",
    "pd.Categorical.from_codes(codes,categorical)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "与显示指定不同，分类变换不认定指定的分类顺序。因此取决于输入数据的顺序，categories数组的顺序会不同。当使用from_codes或其它的构造器时，你可以指定分类一个有意义的顺序："
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-11-13T11:38:53.638701Z",
     "start_time": "2019-11-13T11:38:53.628790Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[foo, bar, bar, baz, foo, foo, baz]\n",
       "Categories (3, object): [foo < bar < baz]"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.Categorical.from_codes(codes, categorical, ordered=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "输出[foo < bar < baz]指明‘foo’位于‘bar’的前面，以此类推。无序的分类实例可以通过as_ordered排序："
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-11-13T11:39:37.647578Z",
     "start_time": "2019-11-13T11:39:37.637639Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[foo, bar, bar, baz, foo, foo, baz]\n",
       "Categories (3, object): [foo < bar < baz]"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.Categorical.from_codes(codes, categorical).as_ordered()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "最后要注意，分类数据不需要字符串，尽管我仅仅展示了字符串的例子。分类数组可以包括任意不可变类型。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 用分类进行计算\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-11-14T02:28:03.979935Z",
     "start_time": "2019-11-14T02:28:03.887517Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([-0.20470766,  0.47894334, -0.51943872, -0.5557303 ,  1.96578057])"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.random.seed(12345)\n",
    "draws = np.random.randn(1000)\n",
    "draws[:5]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "用分位数进行分类"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-11-14T02:28:38.167311Z",
     "start_time": "2019-11-14T02:28:38.128361Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[(-0.684, -0.0101], (-0.0101, 0.63], (-0.684, -0.0101], (-0.684, -0.0101], (0.63, 3.928], ..., (-0.0101, 0.63], (-0.684, -0.0101], (-2.9499999999999997, -0.684], (-0.0101, 0.63], (0.63, 3.928]]\n",
       "Length: 1000\n",
       "Categories (4, interval[float64]): [(-2.9499999999999997, -0.684] < (-0.684, -0.0101] < (-0.0101, 0.63] < (0.63, 3.928]]"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "bins = pd.qcut(draws, 4)\n",
    "bins"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "示例2：用分位数进行分类并起别名"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-11-14T02:29:35.088057Z",
     "start_time": "2019-11-14T02:29:35.068522Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[Q2, Q3, Q2, Q2, Q4, ..., Q3, Q2, Q1, Q3, Q4]\n",
       "Length: 1000\n",
       "Categories (4, object): [Q1 < Q2 < Q3 < Q4]"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "bins = pd.qcut(draws, 4, labels=['Q1','Q2', 'Q3', 'Q4'])\n",
    "bins"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "示例3：根据bins对draws进行分类并聚合"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-11-14T02:32:34.603967Z",
     "start_time": "2019-11-14T02:32:34.555754Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>quartile</th>\n",
       "      <th>count</th>\n",
       "      <th>min</th>\n",
       "      <th>max</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Q1</td>\n",
       "      <td>250</td>\n",
       "      <td>-2.949343</td>\n",
       "      <td>-0.685484</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Q2</td>\n",
       "      <td>250</td>\n",
       "      <td>-0.683066</td>\n",
       "      <td>-0.010115</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Q3</td>\n",
       "      <td>250</td>\n",
       "      <td>-0.010032</td>\n",
       "      <td>0.628894</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Q4</td>\n",
       "      <td>250</td>\n",
       "      <td>0.634238</td>\n",
       "      <td>3.927528</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  quartile  count       min       max\n",
       "0       Q1    250 -2.949343 -0.685484\n",
       "1       Q2    250 -0.683066 -0.010115\n",
       "2       Q3    250 -0.010032  0.628894\n",
       "3       Q4    250  0.634238  3.927528"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "bins = pd.Series(bins, name='quartile')\n",
    "\n",
    "results = (pd.Series(draws).\n",
    "          groupby(bins).\n",
    "          agg(['count', 'min', 'max']).\n",
    "          reset_index())\n",
    "results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-11-14T02:32:54.875182Z",
     "start_time": "2019-11-14T02:32:54.853374Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0    Q1\n",
       "1    Q2\n",
       "2    Q3\n",
       "3    Q4\n",
       "Name: quartile, dtype: category\n",
       "Categories (4, object): [Q1 < Q2 < Q3 < Q4]"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "results['quartile']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 用分类提高性能"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-11-14T02:34:54.866016Z",
     "start_time": "2019-11-14T02:34:53.229605Z"
    }
   },
   "outputs": [],
   "source": [
    "N = 10000000\n",
    "\n",
    "draws = pd.Series(np.random.randn(N))\n",
    "\n",
    "labels = pd.Series(['foo', 'bar', 'baz', 'qux'] * (N // 4))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-11-14T02:35:09.944355Z",
     "start_time": "2019-11-14T02:35:09.089472Z"
    }
   },
   "outputs": [],
   "source": [
    "categories = labels.astype('category')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-11-14T02:35:28.300784Z",
     "start_time": "2019-11-14T02:35:28.292092Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "80000080"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "labels.memory_usage()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-11-14T02:35:47.173601Z",
     "start_time": "2019-11-14T02:35:47.160330Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "10000272"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "categories.memory_usage()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "由上面可以看出使用标签浪费的内存比不使用标签浪费的内存小很多"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "而且将标签转换维分类几乎没有代价，而且是一次性的"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-11-14T02:37:43.276212Z",
     "start_time": "2019-11-14T02:37:36.675017Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "822 ms ± 5.19 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n"
     ]
    }
   ],
   "source": [
    "%timeit _ = labels.astype('category')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 分类方法"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "包含分类数据的Series有一些特殊的方法，类似于Series.str字符串方法。它还提供了方便的分类和编码的使用方法。看下面的Series："
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-11-14T02:38:35.330784Z",
     "start_time": "2019-11-14T02:38:35.308952Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0    a\n",
       "1    b\n",
       "2    c\n",
       "3    d\n",
       "4    a\n",
       "5    b\n",
       "6    c\n",
       "7    d\n",
       "dtype: category\n",
       "Categories (4, object): [a, b, c, d]"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "s = pd.Series(['a', 'b','c','d'] *2)\n",
    "cat_s = s.astype('category')\n",
    "cat_s\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "获取分类Series的分类码"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-11-14T02:39:36.090752Z",
     "start_time": "2019-11-14T02:39:36.071964Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0    0\n",
       "1    1\n",
       "2    2\n",
       "3    3\n",
       "4    0\n",
       "5    1\n",
       "6    2\n",
       "7    3\n",
       "dtype: int8"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cat_s.cat.codes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-11-14T02:39:53.591123Z",
     "start_time": "2019-11-14T02:39:53.560344Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['a', 'b', 'c', 'd'], dtype='object')"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cat_s.cat.categories"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "示例1：重新设置分类"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-11-14T02:41:03.007062Z",
     "start_time": "2019-11-14T02:41:02.986998Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0    a\n",
       "1    b\n",
       "2    c\n",
       "3    d\n",
       "4    a\n",
       "5    b\n",
       "6    c\n",
       "7    d\n",
       "dtype: category\n",
       "Categories (5, object): [a, b, c, d, e]"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "actual_categories = ['a', 'b', 'c', 'd', 'e']\n",
    "cat_s2 = cat_s.cat.set_categories(actual_categories)\n",
    "cat_s2"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "示例2：获取每种分类的item数"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-11-14T02:41:50.989349Z",
     "start_time": "2019-11-14T02:41:50.970913Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "d    2\n",
       "c    2\n",
       "b    2\n",
       "a    2\n",
       "dtype: int64"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cat_s.value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-11-14T02:42:01.184496Z",
     "start_time": "2019-11-14T02:42:01.165588Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "d    2\n",
       "c    2\n",
       "b    2\n",
       "a    2\n",
       "e    0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cat_s2.value_counts()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "在大数据集中，分类经常作为节省内存和高性能的便捷工具。过滤完大DataFrame或Series之后，许多分类可能不会出现在数据中。我们可以使用remove_unused_categories方法删除没看到的分类：\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-11-14T02:43:05.516527Z",
     "start_time": "2019-11-14T02:43:05.435244Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0    a\n",
       "1    b\n",
       "4    a\n",
       "5    b\n",
       "dtype: category\n",
       "Categories (4, object): [a, b, c, d]"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cat_s3 = cat_s[cat_s.isin(['a', 'b'])]\n",
    "cat_s3"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-11-14T02:43:39.559289Z",
     "start_time": "2019-11-14T02:43:39.543366Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0    a\n",
       "1    b\n",
       "4    a\n",
       "5    b\n",
       "dtype: category\n",
       "Categories (2, object): [a, b]"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cat_s3.cat.remove_unused_categories()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "可用分类的方法"
   ]
  },
  {
   "attachments": {
    "%E5%9B%BE%E7%89%87.png": {
     "image/png": "iVBORw0KGgoAAAANSUhEUgAAAyAAAAGLCAYAAAA/NPZsAAAgAElEQVR4nOydTY7cthZGtY7sIxM6G+iJhw08wPCQgdGTAFlERgY0zAoyNxAT7gU00AvoTMzF8A2oH4r/lFSlqtY5gfCeu6ooSuKl7kdeXnYGAAAAAADgSnRHVwAAAAAAAM4DAgQAAAAAAK4GAgQAAAAAAK4GAgQAAAAAAK4GAgQAAAAAAK4GAgQAAAAAAK4GAgQAAAAAAK4GAgQAAAAAAK4GAgQAAAAAAK5GVoB0XcfBwcHBwcHBwcHBwbHqWCVAAOA+wX4BAADgSBAgACcD+wUAAIAjQYAAnAzsFwAAAI4EAQJwMrBfAAAAOBIECMDJwH4BAADgSBAgACcD+wUAAIAjQYAAnAzsFwAAAI4EAQJwMrBfAAAAOBIECMDJwH4BAADgSBAgACcD+wUAAIAjuaoAUVIY0etVvwWAfSjbrzbq+4sJLVWbv//sw7/r3gjxYBSmDQAAABVcT4BoZT6JznRdZzoRcWI2oYzsOiPEg5Hq564lA7w3SvarexG1UyUH+5Uq/nfRR0WIfu7NU/9i1Pf5+Pb1s3nqX4wxxv5/+SV59Ng0AADAu+LKIVjaKDk4N50w1ZMhujeiExlxoU0vOtN10qiVNQM4CzUzIH10sMD5uxwsTfdGFAYVJkHjHeNs6CRgEofEqAEAAN4VuwsQ3cvsaOaT/DLMhIjqEU/9PDg5XWd+80Zfh28gQAAqqRtAiIgNY8w429h1wvTq5/Cd/GDCKECmYpRcCJDg89TvAAAA4F2wuwApjWbWHPF1IrNDJHoVhG2IwSn6Rf5B+AZAhvoZTG168WEQF3qyKSmEDXcUY6jW42xvsRAsBAgAAAA4XCkEaxw13TpDYRfHLkZnV4kZgPOyLoSyxubiMyG1AkSIh8hMKQIEAADgvXH7AkT302LVzJcIwQKoJGm/ug/DISdBkbcxO/O5TYCwBgQAAOAcXFmAhCFS7hFm0Ekthk19DwECUCJpv4MwiDv/2wWI6NWUAYsQLAAAgPNyZQGSOxLiYZG+9zEhMBAgALXU2G8oKLYLkFR4JAIEAADgXFxHgGRTddaIh+E7ojdv5i06e5JahP4kv5inPrapGsA52SZA4jb2KZMNi0XoAAAA4HIdATKGdkQ9idr1Idpo7X6/4dh940OA+2WdAKmxOxahAwAAQJmrCJAxNW88I9Xg2FSIBN0nNiMcHRrxEN2JGQBm1gmQkXnGUntCIkVJgLARIQAAwLm4ggBxNi6L+im1AiQVqjWPzI7iROvXhvoBnIt2AaLN33+O9unYa6MASS1CT4kdQrAAAADeJxcXINMC1JTAyK4PcUiEcfmzK/nZFgBoESD/6P+Wu527dtgoQOKL0NMhmAgQAACA98llBcgoLnJOxPidrJeRcFKioqQ04wJwbuoFiC8W5rTYUpmIAHF3Tp+ZZ0D0WPj878wABAIEAADgfXI5AaKdRauBB6EnZyNwTmJFxb4ziZvI4vVpPwPS8gL4tAqQYMPAUSz4AiRhk7oXRoiHhVDR+tVok5+xRIAAAAC8Ty4iQLSS08xHNLQqu+GZX1jo1Hz7+nkq382cI8TDdHRJ8QNwbsoCxF9XpY2SYxiVM7M4zV48LrJWVYc/5gYRDAIEAADgvbKzANHL0I3kuo5xX49H84v8w/SxzFZeea5T48eUj6LDTeH570tPGk+ACHUzmL0VGk4YZRjWOIdkuTZfZW7ODGlsZnPafJRQSgAAgHfHBWZA7GjpU/+ypV6L8r59/RwuUK3JdKX7eNpegBOzZh8f8XHvvXSsXcfKnQYxxOOO/QgAAADcCtfZiBAAbgbsFwAAAI4EAQJwMrBfAAAAOBIECMDJwH4BAADgSBAgACcD+wUAAIAjQYAAnAzsFwAAAI4EAQJwMrBfAAAAOBIECMDJwH4BAADgSBAgACcD+wUAAIAjQYAAnAzsFwAAAI4EAQJwMrBfAAAAOBIECMDJwH4BAADgSBAgACcD+wUAAIAjWS1AODg4ODg4ODg4ODg41hyrBAgA3CfYLwAAABwJAgTgZGC/AAAAcCQIEICTgf0CAADAkSBAAE4G9gsAAABHggABOBnYLwAAABwJAgTgZGC/AAAAcCQIEICTgf0CAADAkSBAAE4G9gsAAABHggABOBnYLwAAABwJAgTgZLw/+1VGdsKIXh9dEYvujRS/m179jH6sZGd+k8q011Z7v9GmF52RalUtgzp1fp30jdzP94qSTe1AP/fmqX9JfHipNncJbLsV/dqGq4wUv5t/9H/xz7ruuL7gLp4D/QjcBggQgJOxxX6V7IwQD0bd0jtF90Z00uzw/lyw+lqVNJ3oE06GdZDWvOx1L0zXCeP6VroXpmstTPdGeOUo6Tltujei2+Ysaf1qD/XD/P1nb/6SvwbnvQhKXqiN7uzcalte4LAtvqPNt6+fjeg603XCiI99/Lou1OaqUdKIj6nz+2x0eJUM7MD50MimNjY8g9YjVfldnoM2vfhwsed1b/0IvF8QIAAnY739rnNklFzxgk++xEN0L5q+X1nr1U5b8BJefmi6VWJpHDX2y211uIbSPCdE984M0uA0FB1tJSue46MR4sEI8WCe5Bd79C/Os7LX1dw+gsO5p1qZv+Svpuu6yWGvaoPFh73uXmfRdjTfP7N9Pvb+PfUvRTF1mTZXS+t92SZAoqPsFXWJD8Snvp+u48JWInXb5TlcaFDlJvsROC23KUBaO8yLd7DvDO7XqVltv1dqN0lBMbzQtjiq1S/D1deaf5FnHZSV9Vk1ejnUZbzPk+PQ4jRkR3v3Y3Q4m9HKSNHmXBZqsl2AVIg20evGe3uhNldJe/lbBMg8MNA+sBGznz0FyL7PYa1dZ7nFfgROCwIE4GRk7feaTn6yCg0zGpEwgIYT7XitmZH88Vqy58v1X6lRy5G1szV6GhWenSpt1PfEOoPgtFcQIFue7+5l5hzM9PNf5UNW3dtLtrlKCiP17SKh8FyK7/o1IVhbBcilnsPeM2432o/AaUGAbGa/BVwA12CV/W4ICbCC4rH6RdoiQFaPjucrsOpalYz0A4Mj+WberJMS92iy8fNV9yMbF292EFuR+3EFAbI5vK4mTCwjGKZQqOJ9ib0HbCx/zLn99vXzMqxqEEVyXLxceW8v1eZqWTdqvvadOTr6tyZALvcc9py9utl+BE4LAmQzV1jgB7Aj7fYbjpzFFjImaXRUq51OdyS75sVYZaTrrzXnhNg1CYkyck5Iw2h90zOxvzDfvn42XSds/URigXOKFc59eqR75WJgr9/X6keh3Tj9dfNMiOesLoRqiwAZn9Vc98CR30GArG5zDUTPX2SdAJnFoHffojMP89qjxUxEeAWr2l2LANn0HJTcZ4DllvsROC0XESC6l0tnQDwm0tLpabHg+FKS6mdGULR+P1dJGx8clNVwHdHOz+3onMWQ4zmWCzDn7/XiQ3ie6HVpo6Qo39/xBaZfzSfhdZqxcmvqqtVU1nhe/57B7dMqQKKCoEVUXEiA1M1+DKOmleffcq2lUdBkSEKq/OY4ats3WFsvZJ0ZbXlwFsZZKvs3a9fFs+46A+KPRCsjgz7Km5kKZqrG60/0s2Nb8B39WmfLd+IWzmObALGnHt45sRm3rLib67B7m2tibYjQGgFihcL/+n7RLsIZmEidsk78dWZAVj+HPRaj33o/AqdldwGiR+dW+3+Lj3gt07PZhv8kv0Qc79bv5xjTKc5l6WdrpGMn0nodYUcVnsMKDd9parmu2Zny6xV0MEraDiC2CDMQIDV1td9Z5KG/RHw2XJxaAaJ7OYwgJtpQiwBpHWUslO2PICe+VS0+9rjWnBMy21VkL4dI+ePgxrQgueGezRmUwjpr/ToMvIhFPdxQj3nQQuTTZ96cAJn//snvZx1HKbgnQ//5m1T5LQv88y2uv06AVK2JEL191zh1TYU67dnm2ok8I7f8VTNasZmtOfTqzbwtzxmIi9sTINueQ+YeV3AX/QiclovMgIQkwhqihheP82z9fo7UyGl5RDW2iCsuQJKxsZ7T3nJd2ZFhX1QMnU101MP7blVdL5YWEK5NvQCxL6CoSVxwBqRYr/FFmi1zxczHxmutcUKi4Q1B+dZxrbHd5Hn8UCQnRC3maMf6Pv3cJ532eL23EBMgBcc02yfZhbH6uZ9muicnSCvzJL8sn8FiBljGQ0h2ECDFMqNlbxMgdW1uBaveCfNzFeKhLrxIa/MkvwzXqe5WgKx7DlsWot9JPwKn5UoCxG+Y+WwM4ehm6/dzlDJB5AkNLCZAclPM7mct19WYwSIXWrDofGrrOpZPyNW9U7bfMaZ3buu12WySL7tdXj5zvZ7kl51e3Ptda/J3Xj3dtJXDHzaLuZrFwMn1EVqZv//sEztL6+SswL77r+w4A7Ks5bCpm9Nv5X6nVTp+3b/3NyJALt7mWuueQcnO/CL/ML2w4VTtM+gqDMHaNKO6nwC5zHO4wN4zmfMe0Y/AebmIAHFHnRaHF86THpSIhwjVfz9H/aLx8nWkyisvbLMdWMt1lertCZTcPYmUW+NojaEK44688XU9cOuk7Vc7o1WP1jFftePvhu8mmVNdTuu+3Awz1aEd48t8/2utCsOwf1zuqN1wf1IOwqY9A1aGUi7OWRveUT0SvYMAUXKZ0Ur9sOLC+Z3Wr6Z2UCq6UNwbVCsKkMp9QPxzbQrBsn9c3ebSNIYHDe+d0WalWtNuE+ecvNvr7IS+KgTLOV/9c9gWgpXilvoROC+7CxA7EiBMr34uFO9y5uD2BUjddaTKqxU5tyNAWvocrV/Nt6+frRgRj4Rl3Rk5+9W9E6+bC104JATLGUXbYeRw72utd0K8cIzq+5N2lEuOw6rd6LPiIdLf1KzbuZoA8e+V8+/hd/++OA5TxWL0ICXqGgESVDN+DfsLkLVtLkeLs+++Z+qjAOLlxNaHrgwVTjrNF1gD4vy++jlcJPT5lvoRODM7C5C0M3tfIVi115H6bm2neukQrBoBsiUkbVs4GxxDaQZz4uYESEuZjSOhVxYgxjgbddXen8wI45aRyzEsRjY5B979vYgAiTkwlQIktl5j/Lfzmduf59cAhv388nra9gGJiiHxYObuO7JmcqMAWdXmstRms4qLwfnWtWRpis8IjOGA2ZnMaNVSbehyAqTpOVyi77ypfgTOzL4CJNmwVdAog1jIuZDoovLW7+coLkJvuI6UWEnHRysjxe9TnHHLdbUvQq8RIHV1zb0AESD3xb0IECVFmDWmusxbFyDryk/VcbUdaif73/D/f6twHoI+48ZmQJb3w3OA3d8F/79m3VzsehoESOT8y83qwrL2ESCN36mg7LBGUh/H7pXT9vJ1SockjaPzbfojnwTmMgKk/juXeL/eUj8C52bnGZC5s5ma8LDfRpAW0RyYhncacUml4W2/jtBoZ7HilhF0xrGOd9gXpDUN77Ja9QKkqq5TB+Psg+KlLob74D4ESEzst5T5zgRINkRo5cZug/366xpsaGVuM7FIn3dLAsQTEtYxTf3OFScxZ9kpw/t7TICk1xyZxfeiYmjol21oWE7sLOt1pAApLeiPZz9KtNcxw1Jy3zBj0gLEvf+icn1ibgb/FgTIBRag31Q/Amdn/0XosU31dEJ1+5vfZTfgW/H9HJFN9RadVsN1LDcsdF909RsRBnWZ4pLLGxEK8RDfiLBagFTWdcyv79STrFj3R6sAaY37DV7ow0tPqp9G69f8oX6Yb18/D+0sv8D4EgJky7XWZsLxKY0ij2mHF/dVK6O+v5h/Xwo7LcfK06+TrUdHnB07Fx/9rDaJGeddFqG33fugv3XrNw2c2H4tHG2ff/ft6+e5/9Q6vB8Lp00brX4MIiETVjT9bTkDEoQLe3XR+rVC7LhVu0ybayE2ap7bQyLv6I7tK9wYeDhbIED8NK9TspThnZ4Nb1rhjG/JghUrK/Uc9nxGxtxaPwJwxTS80M6+aS4BLLvMgDSie7kUr6lDPBohHhI7WXt1i9nG8EK1qXV3mgGpLqJ2FNSm/n2SX6YXeHRB6LPzAo98PibKEOKhcJ32fIuMftlR5vk37kDHfG2RlJoVg0BtMyAVREbfo3stuJ8PArc1aUkQPjw4Ve5sfOiwRsJzpzS/2mmnflrZ8qJ0ty57trl1xJ+dVj8S3y+PtNvsZKlz2ZC1eaAitmO3337F8nnpwgyrrYTzffu81PcX8ykya7L7c9hx8fnt9iNwdt6XANE2ba4QD8kjGVN+EDmRUd4YEaCdWvu9aQGspOcATh9Yx0I8Ntn69mvVw0i7/zKO578fHeXo7OX8pfznLShpxOCEtYZE6OfIDs6R8ssCJPXMjFktQJyF28MfjfqeFq/Tfe9r+tXBeQrOEf+uVj/CZ53c/GAe6Z/v7eCcRu51/F1wgTa3lqQ9xtiQvMQJr9PDOUvteRyp94WgFL83OPjz8xIf+6DN7fscxr1rqitXqPod9SNwKt6XALlHIutRxjCr5rAygAqwXwAAADgSBMgtMCxwX8SWrxhhAKgB+wUAAIAjQYAAnAzsFwAAAI4EAQJwMrBfAAAAOBIECMDJwH4BAADgSBAgACcD+wUAAIAjQYAAnAzsFwAAAI4EAQJwMrBfAAAAOBIECMDJwH4BAADgSBAgACcD+wUAAIAjQYAAnAzsFwAAAI4EAQJwMrBfAAAAOBIECMDJwH4BAADgSFYLEA4ODg4ODg4ODg4OjjXHKgECAPcJ9gsAAABHggABOBnYLwAAABwJAgTgZGC/AAAAcCQIEICTgf0CAADAkSBAAE4G9gsAAABHggABOBnYLwAAABwJAgTgZGC/AAAAcCQIEICTgf0CAADAkSBAAE4G9gsAAABHggABOBn19quMFL+bf/R/8c+6zoher6iBNtr7dy86I9WKotwayc50Ui3L1mvqZ4zuhenEo1l1eRGUFEaqn/sUBgAAcOcgQABORrX9Kmm6TiSccGVk8rM8uhdBuboXpmtRILo3witDSU8Q6d6IrjO/+aKkglHM7IVWcnVdAAAA3hsIEICTUWu/0RmF+dOkAMlPOtjZjnDmpF3Q+EJG92IudxAfa2do4nXciN4yawQAAPB+QIDkUNJ0nTT7jYPuxK3WC+6COvu1zrJUgxDpWo5M28y03eZZEDPUTfRGG0eAVIoPK2Bar215+Odov1eV9w0AAOAdgQDJcauO/q3WC+6CKvsttrE1IVilmYVZ9LSUqYMZEG3U95eWQpa12BB+FYSBAQAAQAACJMetOvq3Wi+4C8r2a4XC3gLELuzu82sgcutOhpmNi84w6N6IDbblCpD6GZZ1a2kAAADuFQRIjlt19G+1XnAXlOx3dpznNpZ0psWjEeLBCPEw/DshMCKLxvPnr3XKtfn29bPpOmH+kr+aTvRGrXbmR+FVODKzI00zINlF/gAAAO+XiwgQ3cvlSKV4NL2fglIr80ksv9OeplIbJUX5XMbYl73ojdav03lnR0Fb58UZkZTqZ9rR1yr4/lP/snS8cuer+f2aegFUkLdfGwb1v7430hcgC8c7MgOiZNw5b14Qbu3a2kwma9TYhwyiY0yda/9m+5MW335cv5GLvioJjNTnupeL+rSJLAAAgPfF7gJEj86x9v/mvmytk/PUO3HaDSOkww/saKU34jmmuwycACWtUyL8c9i6LNNjWgfoSX6JOPpjJhvn+1rZurieS+F8xd831wugjrT9zqFXb+ZtIUBCcVEnQMaZE9HrQThXzDA4i8pjsypavxopxCTc59M74U+T8BcVqW9nwXMxAfLcT/UR4sF04nHDTA0AAMB9c6UQLG/x6cY4a1tEJp48NkMwOD++c5AuJx4Hn8zU4wuo3Pkqft9aL4BakvartXmSXwYHXO0gQLTpxYe4w56axRtnDscS1A9HqM9rQH6TKkj3G1s8Pjn+ydAwNQmDf/R/xQ0R1wqQoD7YLwAAnJirrQFZOgdjtpu1OwOvyKYTjbfOl2NHYF1HIbdjs/dZ5nzl37fWC6CeOvtVYQhWw+xFuXgZ/W4pFe9CkCw/MH//2Sd2bdeRvUmcWY9pNqJuDUiNAFmXjpeQLAAAOAeXWQPy3A8hEunFm2OoVNcJIz728XUbSUrpOiMOfHTEtVBO8Bv7/SrnJHO+8u9b6wVQzxoBMjF58ut3QjcmLTTW7AUy/LAxhNMYY7S39io3QGDJz3CUf5+a/d1753UAAIBbZncBYkf+hOnVz8WoY+oFq/Wr+fb1sxUj4rHSqT5WgFT5CWvOt7peAPWsFyCO6NgURpme4csJkG2b/JUzWB0mQFaJJwAAgPtlZwGSdpzLI3ylsKqW76ZCsHyHaV0IVlUdV5xvfb0A6tkyAzKuTfr29fP6EfuMw712BkTJzvwi/7AzjKtnEraGYFXMCkUEiJIrZ30AAADulH0FSNKxUAvHIOVktOTQX7cIPRH60LoIPfp9ZaT4fc5skzhf7e9b6wVQy6YQLFOXrjZbcmYwYtVO4sPidKnm///bqsptnAGpmZncIQEHAADAvbPzDIie0sm6KWalEDbzy/hmnxyGed2HfnaciCryaXiDcpLOwbo0vP41RtPwZs5X/D1peOFCbBMg7iyBaFy7ZQqb71WEMPm/GPoN3/ZEF/YNFaVtEiBVszcIEAAAgAssQtfK9OKDkxXn0dgtADwnQfcX2YhQiIf0RoSpF7+/MeC4meHWjQhrz5faiLC1XgAVrBUgfkrbKZHE0C6LgYWx1NRaGfX9xfz70g9tvXK3dP062UZ0nw+nfxEfU9mxgh9tECCV67sQIAAAAJfJggUAt0uLAHkzb+bb18+zMx/sTK7nJBJOVrvFxoHPjhiIOO9j4gohHjLiw55nkV1vFORJ/LqVBEJKgNhynuSX5I7u9hoywmIQWkpmQkcBAABOwu0JEG1T+ArxkDwWO6gDQBNV9uus59JKGvGxHM40zkoEDrzu0zOTLShpxCBwWncR1899Rb9hw0WTSTQi4moo3Wj1w6jvmfLHXeBXzfQCAAC8L25PgADARcF+AQAA4EgQIAAnA/sFAACAI0GAAJwM7BcAAACOBAECcDKwXwAAADgSBAjAycB+AQAA4EgQIAAnA/sFAACAI0GAAJwM7BcAAACOBAECcDKwXwAAADgSBAjAycB+AQAA4EgQIAAnA/sFAACAI0GAAJwM7BcAAACOBAECcDKwXwAAADgSBAjAycB+AQAA4EhWCxAODg4ODg4ODg4ODo41xyoBAgD3CfYLAAAAR4IAATgZ2C8AAAAcCQIE4GRgvwAAAHAkCBCAk4H9AgAAwJEgQABOBvYLAAAAR4IAATgZ2C8AAAAcCQIE4GRgvwAAAHAkCBCAk4H9AgAAwJEgQABOBvYLAAAAR4IAATgZ9farjBS/m3/0f/HPus6IXq+ogTba+3cvOiPViqLcGsnOdFIty9Zr6rc/Sgoj1c+jqwEAAHATIEAATka1/Sppuk6YuMZQRiY/y6N7EZSre2G6FgWieyO8MpT0BJHujeg685svSg5AK3kzdQEAADgaBAjAyai13+iMwvxpUoDkJx3sbEc4c9IuaHwho3sxlzuIj5oZGiU703WthzTNEzZ6y6wRAADA+wEBcgRKrnNgLs2t1gt2pc5+rbMs1RoHPdOGMm2seRbEDHUTvdHGESAN4iNJZIaluj7NYmaDqAEAALhDECBHcKuO/q3WC3alyn6LbWFNCFZq9sMts3UtiJ5mXOYZEG3U95eWQiLF9kassIUgDAwAAAACECBHcKuO/q3WC3albL9WKOwtQHQvptmKdLGZdSfDzMZVZhh2ECA2RKymXuvW0gAAANwrCJAjuFVH/1brBbtSst/ZcZ7bQtKZFo9GiAcjxMPw74TAaAhpii1Sz3zbfPv62XSdMH/JX00neqP2cOZX2kLTDEh2kT8AAMD75SICRPdyOVIpHk3vp6DUynwSy++0pKlMj6aGYR7jd9/Mm3VS3HrFvSWjpChfg5K2Dvp1upal86GX5+uGVJwx50ar4LtP/cvy+krnqymjtV7w7sjbrw2D+l/fG+kLkEVsVGQGRMn4Go7mNRnW/mz7zmSNGvuQQXRYO38c/mb7k9W+/WhrzT+LX6fu5aI+bSILAADgfbG7ANGjE6v9v7kvW+vkPPVOnHbjos8WAWKdiUcjhXBExOjk+Occwk+8kdQxjWay3KAi9hqXaTftOZ/kF8/RH7PjON/VytbDdegqzlcso6le8B5J2+8cevVm3hYCJBQXdQJknDkRvR4EbkVIkrOoPDarovWrkUJMAns+vRP+NIl0kU59W1uf3OFdb1KAPPdTfYR4MJ143GemBgAA4A65UgiWJwpWxlcvSmwVINHFrWFazGycuj9DMJQbH/HM1y8Ib0mOHDtOXul8FWW01AveJ0n71do8yS+DnagdBIg2vfgQn/lIzbZ5Mw9a/XAE9bwG5DepgnS/Y9rgxSWNjn/TbMbch7Rm5iqFYM1CBDsDAIDzcrU1IEvnYMx2s35n4HYBEnvh+99tzNKTjOHOl2NHdsf65HaB9j4rnK9cRku94L1Sn4a3Yg1IYvaiXHw8xKnk8C8EyfID8/effWLXdt22Iboj2tcKkHXpeAnJAgCAc3CZNSDP/RAikQ5XGEOauk4Y8bEP11cUuIwAKaUB9b6fLLdQzuJ39rs5x6T2fOUyWuoF75U1AmRCu7ay3mFOOfZr9gIZfrhq345k3dwwsOr65AYC3HqG9zU2ewMAAPBe2V2A2JE/u9ZCa+/vMYdDv5pvXz9bMSIeq53f9yZAqnyPtedbVS94r6wXII7o2BRGmZ6Jyzn82zb5i6/ZiF+3l0r30gJkR/EEAABwD+wsQNIObnmErxT+5H07KUDCdR2XDcGqKTdSdy8Eq+q6V55vXb3gvbJlBmS0u29fP68fsc843GtnQJTszC/yDzsTuGEmwbeBtvpUzApFBIiSK2d9AAAA7pR9BUjSsVALxyD1Um/PoZ9YxLpagKxZhB532JVMxcMnFqGnxJT4fc6WkzlfbRDCUxIAACAASURBVBkt9YL3yaYQLDPPRKz1mXODEat2Eh8Wp0s1///fVodxLa8rJ0CC9Sg1M4g7JOAAAAC4d3aeAdFT2lc3FawUwmZ+GV/k04t+Xvehn8OXf55Z1EzzF8+9ER9785f8dbUAKaXhXSYCyjkc7Wl4/fsWTcNbOF+xDNLwnp5tAmQUqnOoZRPZzfcqQpj8Xwz9hm8nogttOF9QfK+StADRgW1VzZYgQAAAAC6wCF0r04sPTlYcu9lfMOqp+00bEcbKEB+twxGMojaHLoUbEQrxEN+IMOdM+BsDjpsZbtmIsOV8qY0IW+oF7461AsRPaTslkhjaTzEAMDY7qZVR31/Mvy/90CYrd0vXr1Mbju7z4fQN4mMqO9Zcr9Ssybi54T/6P6P163z08b2NiuIJAQIAAHCZLFgAcLu0CJA382a+ff08O/PBzuR6TiLhZLVbzCk+O2Igoi7GxBVCPGTEhz3PIrveKJyT+HXzBMK4k3qiXvY73kCJc53ugIm9hvzggPr+Ytd7rNhhHQAA4D1xewJE2xS+Qjwkj8UO6gDQRJX9unthKDnNLmZ/MsxKBLMAuo/PILaipBGDwGndRVw/95F+w26UuGU/orEcrX4Y9T3TL427rq+Z6QUAAHhn3J4AAYCLgv0CAADAkSBAAE4G9gsAAABHggABOBnYLwAAABwJAgTgZGC/AAAAcCQIEICTgf0CAADAkSBAAE4G9gsAAABHggABOBnYLwAAABwJAgTgZGC/AAAAcCQIEICTgf0CAADAkSBAAE4G9gsAAABHggABOBnYLwAAABwJAgTgZGC/AAAAcCSrBQgHBwcHBwcHBwcHB8eaY5UAAYD7BPsFAACAI0GAAJwM7BcAAACOBAECcDKwXwAAADgSBAjAycB+AQAA4EgQIAAnA/sFAACAI0GAAJwM7BcAAACOBAECcDKwXwAAADgSBAjAycB+AQAA4EgQIAAnA/sFAACAI0GAAJwM7BcAAACOBAECcDL2sV9tevHBPPUvO5RljJLCiI+90Svqob1/96IzUu1Rp850Ui3L1+01hGujjZLCSPWz+L0n+WWnNnywPejeSPG76RPXrGRnfvPbcqG8VFlGSdOJR6MwBQDYAAIE4GTk7VcbrV/toX6Yb18/m0+iM123dOp1L0zXybITonsjOvv7xbFQCMrIrjOdaBcgth7C9Nr7W6sC0b0RXjlKdkYsCzaia3TktqB78yS/7Hiu4T43H8v7ElbTOsvltqDM33/2JvtkdG+kEGnnt5bhWeWbQY1YvRN7UDLzfVtei0ko2aWvJ3uuFNrT7vber7F5AHgfIEAATkbWfn0HSTwa9f3FOmDjd5RMOqvC91R1b0QnZ6cz4eh3ojd/yV8bhYN1YoJzGmVkwWmOluaJGd2LuezhvoTnuiQpJy0tJPK3L7wvSnq/UTLiDOfu5VDH6Imt8/7t6+dFm8qN7E+OfO4yalGyIAoqBMid2EMglpcfNt5T275ErwchUn+kqjtek/b+dl17AoBbAgEC16X5ZQh7U2u/sZmE0Un3HY2YgzH8YOFwjWFNzg8dp986hL/VipBMW1o1C2KW1zEJkEPEx1QjKzYWsy4xUVAzmn8BATI8v3/0f6YXH4wQD+EIv3g0T/1LRfSaIygzTn2L85s/Z1u43u3aQ/4ZNTv6hT56nW2FszA3K0D86+edBXARECAAJ2O1ANG9EeLB9OqnkWJ2umJhUIvfTC9vz1GKhskoG+JSDM1IzX7M5bSGnYzljk7rPAOijfq+T2z/KnS/dEL9UXT7x0oBsm8I1sKB1r156l+mGQL7J1HvZEavq/D9xWxK5vqkMtNsTel6Ezfx9uwhcz3j91MhX12XdKoDURT5fI1wWM5ulez3QBAgAFcBAXLX7LfgFs7DlhmQCSWN6IR5kl/yawQch2tRXnZWQdvwk07YkfNU3UoiZTGanKrbGoc878BdnGgM/pa+QBkpRPsC6tL9NW0CZFp3UHv+QIAkvuPNKMz3yP93XrTeoj0EM1hDXTrRmzfzlg6PS9270SaCWZl1os2rmJHi9+GcN/zuQoAAXAUEyF2zdpQXzkzafgsj5F5DUzIz0jsyOFz/vlhH8B/93/C7zog+33C1ktYZ8p3tSNx8+vQVdVz+wnz7+tl0nbBOn6hYXH1l4iPUo0P3c140XV/iin5kbCs7CZDJyd1TgPij7AUBoqQ3A3P79pATIKNoSc7EBPfOmVHxBUhG7Nc846lNOodtcz/Dz9QP84/+L1veRUGAAFyF/QXI2Fnp1ylbyDKTjBo6xnl6PxjldEZw3O/OHaaeOu0xxjje/9nv+YsI3QwrTYv3aupeg7Yjjm45fspI3ctsva1jlZh2b6mrVqYXH8LzJDvd8j2d7l2sHaTKramvVlNZ43nLqTbBp3YAIRmKodWwOHpMxemnwnW/ax2uf/R/k8Pxl/zVOh6xNuwcY/jTIo6/eT2GXjh4+dHyOdxF6XGW5XH4m21r19EinuO7sOu0+Fr2Gb6NrQ2/mu1xPqeenn9JCNYJEFveL/KPhQBItQ85f6F6QbuumgHJc4v2UJoBSYYORkTFKNaDxe+rBUi6zU39uHi064acI/AZMueN+QkxfyA9YxoJBUOAAFyFCwmQR5tKMbD2MbuGWjjKwTTxUMaT/DI7tsP3fpF/eDnex0wd8ZhofwRzHEWaOpxM57J84VTWvUhYjn5exv7qSPYWHQ13SI1c1tbVfm+ZVtQ6bDaUYOU9NSbdDqL3u6a+9juLMJGGkXCYqRIgU3iSkxXquZ+E89xmxlHTRLYhN+Y997wqwmlGB61pkbKzoDxw5sdy9etwXcswJHdwYhbS4nppeG0l5muI2dkOtC0qHp+3GJzoObwpVXapvmP/bWcFPMdvUS9PMOTajNtfDu1Oqp/rBcgN2oO9zLQAcUVrEF7nfcdtW0F7qLC15jaZFDUNa0Pcd8zCH0ik5kaAANwUlxEgiQ4p+aLzO+KhjGjHGis7sngvGyO+6FDyTvxiYWFN3QukRtFKC//iHXO87rV1LXXKfqdbf09N+llFOvOq+rYuUIUkNQJkdtiXo9zfvn5ehEcU06Z6zy37rIvhNB/ijknKQQicrB+RUK5udiCDphrapH7uKxfJ70TkGrYXuRx5Hmc4/JFoIR6ioV7fvn4e2sO8iLo1XevUv7t9t2/jKwWIHh1m8WiUcYVkeRF6dv3HTdmDpUaARMMQYwJA61ms7xyCFat3KnVztShM+QlOGmG3jggQgNviQgIk5pDnOpYwFjceppNy9n1HvC1DTrQzXHS6DXXPsi3zR1jPmACprWu+LuGLtDHrUKodBM+2tr5qClmAbZQFiHUsp1CMxLNMh4z4mX38fQ8Str1qJ3STdJBqRvYDUTJ/YP7+s0/Eovubql2QxbXVhFGVB0Oa1mVk7195v5VNWbBaBcgUxinM/3o3wYCbeelDONMSO3fkOm/RHpLCL7LnxuJvGVFxeQGSW3PUsB4pEybsv6sQIAC3x4UESMxYyy/PYlhUtQApdWJepxO8BPxOqaHuWeo7V3d6f3EUBUhtXQt1CZ5B4z3NjUq3LPR07u20CLOzOy9v3i35pBQFyOD4uY7It6+fw1mriC0GIjmwLevM2YXe9SFUOVJCY+1eIHO9byC8L+r8xR3/2usdn11sxmNxBP1NULnbEiDu7Mz42Rha9NGuFZDTniURBzWT4epW7aEqBGs6j9c/twiQXd5/TvmiN/o5EhpW0aaW9UKAANwrVxcgm0Y2LiVAYi+2xXn2yjZVV44d1bJxrf6Cw1oBUq7rbQmQlnu72F15CLOAevICZH4ecacsHmNtfxoZGY45lbvOgKRn5koO+ZqwofyAwAWont2pd9zuZQZk0yL0xWeRmY8gDDd1HbdtD/UCxBNJh82AOP19LKFEi/BHgADcNVcUII2Ly64YgmWLnjuo8EWz16ZJNeWkHfKWEKzazDOXDcGqESBb7u0Nb2Z1w+Ts198JfDl6O8xWicdwFDm6iZqZHC7r/HmjpcMGe3r83uhwuX8vkc0KtX4GREmb8EJeQ2TkK+LsFaE8u/lQXqMW4W5mQGq+XyVAxuouHenxGQdZnxY/uW17aBEgxjgbah4kQOIzQrlohGxhOwiQcL0IAgTgOlxRgBQ6AfH7nDVkswDJnStVvlpMz0cXdtfUvUBxEXrhGssCpL6uQVzwXMJOi9BrBEhdfVMO1tpdec9M0n61tpnnJl8gvOeL/SUGB8ymT008h8GheDNv3poK28biDtfc1qti4DPO46q2ESyM7pY7kV+TUYAoGekXhg0Eoxnr0tzLDEjV96sFSDqJR/Le3YE9tAmQuu9cKgQruUBfO2mLa+q+qFflYGf2fYQAATiCqwoQ14l2O+1oGt6NAmRyohMpY1MzDL/IPxIvtcq6l5hGfFJpePVU5uLFJoTNvhMRIGHnX1lXx7lafu/DqjS8y6rVC5Cq+k5OobMXynNilBGylBehW8qj6nP4STI5QNSp1PG26C0oLs4+JBNezOdobRtjm/L7o3EDuOxAQ7Y+65jDkOKJPab9kKY9KGrLPNkMSOTZaMexrplxu0V7uJoA2boRYeWGoNlBLp+maIvwHaOf7T3+S/6KAAE4gCsLEGOqNyLcLECMGV8I7qZ5QjykFy9nUghX172GyIZ6izrFNgfU8dHe5OZjDRsRBnUZY4QrNyKM3tMmAVJZX92zEeEObBcg2lmDU3DKg6w/Kp7GNupMpgT27Dz6m5yq7y/m35fCLtCxaurXqf1FnVGn7YmPiexYuwuQcY+jyILzIR3wWNdpsbV4KO4ifZkZkMaNDmNO5qUESBDmo5btY2qTkQ1VF0Xenj3UZsGqv5adBcg4kNSwPql6QK813Nt7f4iP9lkFM6UIEICrsL8AgXdB00gU3BVrBIjWr+bvP53MbAVnzSkkcLie5JfFyK76/mJHjqNO6XJX6WkfjoQwGR12IR4KDo91GheZ5orX5Dianbcgejr/3pmz5pS/Wv2wo7XJGQ+vfqIf7nt5/4t1C+4bMhZVXWqNAFHmqX8x6vuLFRGpPmp04McNJEW/EJmhUJhnk4R4iIqIW7SH+hkQ2zae5JfpHmTX/41Zv5rbyZzyeG6LuTZir9+tFzPaAOcAAbIX2r6MciENYcrBY8mJjPLGiHCv1AsQ6WUSmkcNq9G9XceT+Xwcua9yZHWfn8VsYVhX0XxNxiRTiC6cyZ3Rvaia4TDG1q/1mtrZKzvgQOUMSJXIHNrJ6AzPbfhDdtZUK5mcVbk9e9DmL/lr5Hri+9S4oXfrZnoaadzLRIgHZrQBTgQC5MxE1qPMI4FMOb9XsF8AAAA4EgTI2RkWuC8ymqwYEYb7AfsFAACAI0GAAJwM7BcAAACOBAECcDKwXwAAADgSBAjAycB+AQAA4EgQIAAnA/sFAACAI0GAAJwM7BcAAACOBAECcDKwXwAAADgSBAjAycB+AQAA4EgQIAAnA/sFAACAI0GAAJwM7BcAAACOBAECcDKwXwAAADiS1QKEg4ODg4ODg4ODg4NjzbFKgADAfYL9AgAAwJEgQABOBvYLAAAAR4IAATgZ2C8AAAAcCQIE4GRgvwAAAHAkCBCAk4H9AgAAwJEgQABOBvYLAAAAR4IAATgZ2C8AAAAcCQIE4GRgvwAAAHAkCBCAk4H9AgAAwJEgQABOBvYLAAAAR4IAATgZ+9ivNr34YJ76lx3KMkZJYcTH3uhdStuKMlL8bvrbqMz5UNII8ZC4/7XtTptedOYX+cdF25SSnenEynare9vO1M9k2b9J1Vi29r5v74NUayoY1qfz66Mraqf75DUaJU0nHo3C1gBOBwIE4GTk7VcbrV/toX6Yb18/m0+iM123dGJ0L0zXybLjoHsjOvv7xbHwiJSRXbsjp3srWsp1UObvP3tT74Np85f81XRdZ8RNqhB7v1rqpvVr4u9zecEzmg7ZcO+soyrEQ5NTqZWcRYXujXDOqaRwrlUZ2Ykqcaik53in2mKyXVbUuxfNv3EqmGnz9pm0Fm3tcnl/VtVR90Z45SjptbnhfpZEkpJduq/I3gMAeM8gQABORtZ+fSdNPBr1/cUKkvE7SiYduMAp9pzJlGPTid46/dWOkh3ZjX/fiqhvXz8vrqV1hkUruax77ru9yDu2VUeto58Y1dbK9Opn5BozAkMqk3Xq/edXZJ3jvDiP7udnpeTSeW1wWOMCJH0taxz1LQIkcOiXHzYLv7FdhGXWi7ZFaZ6Y0b0jBId+oiyCZ7FshUj9scesDQDcLggQuAyrXqAn5or3q9Z+Y87V6JT4zkEyFCUYzfZEg5KOk2MdqN9qPI/hd//o/0wvPhghHsLRbfFonvqXbJTIeuHQ/qxio9Ply4yJrIQAqWlDUSfcc1C1P8rdcK0b2vHkkA8C5M28BWKmxeG/bQGSFwVZcZIsMn3v19bTtetJgFSLj3ydttQLAO4fBAhcBgTIzbJagOjexuarn0aKWYRkHeuF0+c7utaRWfofyoZ8FUa5F4657s1T/zLN1Ix1anXgtD/a7p9vU6jIMAuxwsENr+OSAsSbWYp9vyaUqXBEn40e1i84MyDL0LG29Qy7CZCK6xXiIXqElznc31g5Y/vKni9V/9Tsx3Q31s1MGT3p0bktaqO+1639igvo5ee3GeYIAJcGAQKX4VABst/Cy/fIlhmQCSWN6IR5kl/yo/qO07coLzuKOq7BEHYGI3Lu0kzCGgEy/q4Tj6HDvXGB/BQHv6I+RwiQZahNZb2bw7XGn1XMQol+EIhpx/zNvKWd+06af19WrgHJXFfaRuxi+VgTDISR/aPpRD9fQ6oeiXZo222hjZbsZrOwTAjVYMZzxTMAgHcHAgQuw6ECZO1o3zlI229hMbJ3Q5WsCCkanDfr/NmQKfu7zog+/4DsGgx/5mGs41YBUlp4vWEkPziVXO1YHSVAprKrRUU4Ar8m5Mx1UBeLm3XEpgt1q5kBUbJibdAVBcgovJOzibG6RtZVpS+l9Zlo8+3rZ9N1wtZNVCR9GH43iUFfgGSE0tqBAwC4P/YXIGMHo1+n7DnLzBlqyjDTDY5EMMrpjAa53507Xz05MXZ07DGZslFJESyqdVMCNi0ErKl7gfRIVeIFHrkPsettKddoNT2bsTzpp0msvlYdfE+qn+sEiFZGChGWtbhOmX2e0dFU977UXpdWphcfwvNEr6vcziZS9pG6XzX1rXmeDrUDCMnwCa2sczGlz/RTf7rftc7bP/q/wYH8af6SvxqpfhZHvsdwj3lJgp7OW3K4bseRcYTOeC+VrFvnYvYTIKkFwLbs7QIk2v+0ZjgaRsz/19tQv09OG1tTfrUAKbWTK8+AJMObYtfbsh5jqJc7AJAXXnM45PwMHqfnIqMJD8ZqiVm0IEAAIMKFBMijkSLmIIwZMZYjW8GU81DGk/wyO3HD936RfxglXcd0zK4ROoW903nOp5PLDrv00p7qVVn3Ak1Cwb2Xi+tNpFqsKtdexyKPfjCCVnut9nvLNIy2fjY0pz17jntO/bxcIxCL0dfRsILUDMglrquynU1FJ+wj2g5r6lvzPJdUCZApHMPJgvPcTwJxvjfjSGcizaaf3ah1dHf+wnAeMYiZ2WGOV7/FkRnTAD8mY/nH45OI9TWFOo/hQ8Mzm4RXxR4PrZmDts+AxNbp5NZN2D07oqPqDQJktPV5kbMTtjeW7bWR0gJm19HX4+9vXIDMdt6He5143xnb0TyAUdE+nAXlweDMWK5+HexcLOrg3qt5AEUEaXjdvi+4PxX1RIAAnIPLCJBEJ5Jf5Oe8wIYy4iN8qVSfkWwpqZffwtnLO6uLhbY1dS/QLECyjnQ4W1Ist2JEs/ZaS+dsESCp0fbSIsbofUs800tcV307c/+WECbed6vquyLuvkaAzA7Kcrbh29fP5h/9n/e9zPm9+mWvqSBAvn39HDjJ7Q56zJ4a1gxVO9VeW3EEyFhOGF4WL2fal8U5Vq9x2l2AiMQ9NdX3auFIm3C9zLwQfVm3knhQ0grzT+O9ql7jEFvLcF0BEg2VWnzHlh+9/tSgmn8O9SM6mzINMARdVNgX6+c+njRiSCgQFSDMgACAuZgAiTnkuZe891kyHCXl7PsOZ1tGkKiTG3T2lXUv0C5AYi++dLhW7QxIOkSn9lrz97jomBaup4Xw+cUEyCWua0XmmZR9BM+6tr6l5xlSFiDW0ZvCJxJ1TodQZRzYlDPXtNC7vK9BmyMzz66UZkCqRMMUEuftXxEq4nx5Wi3Enl/fuN87bLqY9vDyAmR41v/rhzKSzve4NmC2vVoxGAspm8KOsgM6rr0V+t1xptAN44w4v1ExEJS1vwDJzU4svuf+rVb8Jr5Xk/I2ECXzB+bvP/tke0ylukaAAECKCwmQWGddXvRZDIuqFiClRcilWYG4077HtPHxAmSeIu86u5P0cq1C7bUW7nHTGpD6ReNuGNDiKAqQS1xXYzsLfp8qt6W+pecZUhQgg7PsOg/fvn4OZ2cithiIwcC2rLM7Oa+lI9qmLyNA9pgBmRfOe5m0YgLEVjQtQpLnygmQwoxYQoCMomWM85/D5qw4cHoPZ9T70T7HTKa0mhmedOaqcEZi6ueeU0JWTQu5gxTPQX0qn/vq7FDxNloVgmX/uJzpbplRilzUpj03Gmf6k+ckBAsABq4uQKpf8tcUIP6LKLEmYo+sSrcgQKZP3N2iJ4ep9lqvL0DsyKFdD6O19/dKAbLvdV1egLS0ufjzDMkLkPm8cYclvgZpvgbv777Dm7Ptw2ZA5sxa62ZA9GLXdT8m3p4iIUBsZePlXliA6Gd/nYG9D//ry2FX0zXmrqt1EXrkuaZs+1Ni7d04a6B06OjrXphf5B8FwZN671x5DYhbvjs7VSnomkKgnXq1i6zcQBAzIACQ5ooCpCHM5sohWLbo2YEPO+ptIULLS0gJhXBdxz4CJFJutrzaa712CFbaIW8Jwdr3utaGYNUIkC1tLv/bnP36Ox8vZzPmxdp2B3J/FjHyfBZpeL0RTt3PjqwrQNy/x2u5swCZfmW0+jGEL9lzRNvb91jWtD6x8dz4o4yjPvy+fhH3BgGillnkwrS547OqtN0dBUh048tk6F95NH7p6I82oZLCwD7zeH8bfeaxOlfQIkAW4Wk19zMzU7FlBkRJmwCmdTNNBAgApLiiACk4yeL3Zbz0JgGSO1eqfOtwjI5VdAFzTd1LZB3Q9QKkttzUS2iR4aTyWtO7Q19gEXrh2ZcFyGWua90i9BoBUlffmufpk7RfrW3muUkfhGVr/TrXZxAkv8g/0qlAB4f4zbx5MeT2XsYFyPxM41ewrwBx012HM6H++hq7vqM2je78s4IASf1mDwEyhCVNwiOWItoVkFERk4jz302AKGdmwq5Bic9yJPaXiJ4+nNW2/XurAElzeQHS9p1c0o7Vu44HbaO+/ROCBQApripAXGfRdWKiaXg3CpDpRZVIj5oaSf9F/pEIBamse5GwHP1sna+/5K/rBUhtuRGnyk93W3+tO6bhnfLZp9Lw6un8CwdYiIijkpr1qbwu5yW7/N6H5jS8UUejUoBU1bfqeS4prgEZyyk6V3M4VnIRfMKZjd5z1+50TFiO7ClAwgQEy8nPPuJs1cwq+j/ZU4CooZ4/l9mx1I8hFCx2vz9EMxvZj70dq6MiJtH37hqCZevqhrP5e1WMGbLGfWVyjrDr6M+OeTo06u4FSDIBjDFZ0Zph7EtSGdxKA2/MgABAiisLEGNMzcZquwgQY0YHaRFyIB7Si3QzKYSr616D7hcbx4mPc8zyegHSUK73vU0bEfrfy27YV7ov4YZ6i2cV2xxQx0f9lhsWrthMMlYXbRLX1dDOmgRIZX1rnqfDdgHiOIklJyTIgqUSaTtja0ByQtJNG9sQox5bZzEPkxcyK9kwqzk7VINDt0KAxDfVjO9V8iS/mCf5ZWgr9XYX3W3emS0YxU1yVnC4rtb1A7k+1l3g/skLDQ0zrKVmKx1Hf+GY34YAqWqbDefSsXeXVkZ9fzH/vhR2WI+Vp1+nficaDun0OeJjKjsWAgQA0uwvQADeMdmQqzthjQDR+tX8/aeTgSy103tYSCBAnuSXxUyH+v5iZ1Ji91XHdlkvz4BUEQxoWCc32ABuqEcsLWqt77kpA1ELtTuXP88OZMzB9AX1vOt98MUdrktPYiNWn7mukWce2zBTv5p/X/opPfVf8ldnpqSUdeuWZkDsfZmFZSjc3OcYc9zHxB3Z9UnOuRYZBos2vpyt6iKCfLw/6ZTd+z0LALgvECB7oa1zlsugE3Vs3jt3eF9yIqO8MeLtUy9ApDMrMcy6fSyHXXiF2PUqmc/FOGNUXe4+WenCLFDGGCWXs0nerFbTtbvnujnhOoRKbrS93a5LyWTb0krmM6RptfzdOBsQ/U1+/Uy2rUar3dofaPOX/DUyQxlfYzMuuE/OqI7JDxr2AUqipBFDKu/Wdh61JXNF4Q0AdwcCBMAnsh5lXu9w/6Ny2C8AAAAcCQIEIMawwH0xAr5iZPAWwX4BAADgSBAgACcD+wUAAIAjQYAAnAzsFwAAAI4EAQJwMrBfAAAAOBIECMDJwH4BAADgSBAgACcD+wUAAIAjQYAAnAzsFwAAAI4EAQJwMrBfAAAAOBIECMDJwH4BAADgSBAgACcD+wUAAIAjQYAAnAzsFwAAAI5ktQDh4ODg4ODg4ODg4OBYc6wSIABwn2C/AAAAcCQIEICTgf0CAADAkSBAAE4G9gsAAABHggABOBnYLwAAABwJAgTgZGC/AAAAcCQIEICTgf0CAADAkSBAAE4G9gsAAABHggABOBnYLwAAABwJAgTgZGC/AAAAcCQIEICTgf0CAADAkSBAAE7Ge7RfJYV56l+MXluA7o0Uv5te/UyU35nfpKorX8n67xpj9HNvnvqXmm96ZWrTi85IVXmiDEp2pvPrrCuuYM/7FhZu/pK/bnuui7oII/q1dRlq1EsjPvYmfsuV76EbzwAAIABJREFUkV1nRF9z35T5+8/eqMhXdS/sOVZV9AbbyK2jeyPEg5F+G9bKPMkvybbdfhr7XNffMWWkEGE922tym21k775ESSPEg4mboza9+FDZ78KlQIAAnIw6+7XOVNd6ZN5iSkZewLo3otv+UtVKGtF1phOP6xw3JU0nUs6BvRfVL2g93Lvcy1Jr8+3rZ1vnrs7h1L0wXScWL1Tdi+w9TxRkhFeOkp7jrHsjuooX/p73LVZ21+3mJOh+YxsZ65S8XuvU1OiP2DOYy1h/z26yjdw49p7JsE1E7oEeBhdWnMX0olKc5koZ+rkt9/xm28gOfYlWcu4vdG9EJ6fBAjsAMZaujIzaH1wTBAjAyagXIG0Oku5F9gUbvJzsjxYvienP+nUY7auoqle3NS/5aN3mD62D0lQVZaT4PbyuXlihJh7NU//S4Ainrm3di9R3QhbPbnAaau7j7vdt/vEk4nZFK/NpiyOoZKZOrQIkcm8G0eUfdfW9zTayB5PdbDoi99t1onVvnuQXxwFWRk6/0UZJWwchHpZ2O9yL/HPb0eEdZmzWWcbttpFd+hLXrnQ/D3iNv58eR07sBIWaXjS0s+pyAQEC12eTYwJbuV0Boo1+7s2nobMX4iEIv1FyZwcke62ZeodfKp5f9LrxxeeXH6//qtFLM9zLoS7Ts2tyGna4b1EaXviR6967jSSdX+fexcvKOHRRARK5ny195U22kesRG9kvsWyjfgidFSDjTGVbWJzXT2aFa1CrdbPPxX7O3HAb2a8vmb47CJA38xbMnrRda8OsZNNzBgQIwMm4PQHSTaKjE4+mVz+TDnrpHEmiDl/G0R1FQmZ0c5WIXiVASjM7a0Od9BSePd9XbdT3UsjTJe/bUHbhHrlOT+yzdc6xO+KdqF3WcWkYRfbbY3RmpiUc69bayLVpnzHTvQjb0CKsaBAkq9bjuM+uQlDX1HvRd6wJ1bvFNnKBvkTr+XfDDIjWr8E5668TAXIpECB3zX6Lx+A8XFOApEej57AH0Qnzj/4vXg2tF2sY9xUgtn7BtQwv+jfzZl+O8YuNLCi16zrC8AxnjcsKARJ1lHyUrBhx329Udd/7Nnz0PNRxKCO1drV0Py4tQPIOXO73qZFt2/7V9+VsX9Vzb/nuldtIviplkdmC7Wdazm9nrX6Rf5gn+cU8yS9GiAcjxMPk/NqR87iDW8bpJ2OzDsk1QGE5U7KCpACx4Z6lsm61jezVl1SF6Yne6OxsdawNIUAuBQLkrtm4yBNOSb0AaX/B7LUGxPnR4oU4ZjJqZoUA+Uv+mn4ZJxzpaUGr+2/3BK0CpNpRWROCYgVT1wl7raJ+pHf3+zY4BeNi1fFa/OQEZSdqzpzVzJiFJwiBSovof/R/RTuJ9s9Be6yxt8w9vcE2UqrHbovXx2fU8iJUchIf6vuLUd9fzD/6P6P167D+7Pfh2SZC8oo2PDqtP+OZ0Ur9nnuuTtrMUJkZkOJzveE2sndfMv1+sJtFO9MRn6n4LBrXgOCQVbO/ABmNRL/OsdzLNBJDo5o7VT/OezTwN/O2/K54jDbEKbuJ+z0/q46jqN0y5wY8LzJLnqui7lVom07PLSd80eavKar23U6xtq5amV58CM+TjBW196n2fkfbQXREqKK+Q5iCe97tKQnPR70AuVYIlm3/48vfPT4lwnGa4vwLo+W50bdkGEFGSCg52HPsxVZwaJcROK1x1M5C2VK62dGWBmfB9rmPw9+sXZVCoPa+b8swCbPI+DNeW4vTWttGqsssZGyzo/qPdQ5IVICUZk8i9njDbeQ6OMJtvO+rM1XFyl7Ozmr9Wplq2faT/+v7aDKKmoEI7c9GFEKwksLhxtvI7n3JcL3/6+1i/bEuc/283xSfBTMgl+JCAuTRSBFzXsZFXktFGkyxuWVMnf1oFGFaPD+FXmC4TpmLvN7DuX+Rf8xOw3Qufzqusu5FwnLG0INpNKP2mpIzILV1td9bvoDtfX6SXyICZI7P9usWdHC5dhAIkJr62u8sRjYbRnVg5nABotWi/Xz7+nkpLB2HvCVv/ppFlLmX39wWI/t0eN+pcnbH6X/hhQ1E6jwOMEwL1xuE1pxpKzLTMGQXGwW+ex/GZzcPGqQd/r3uW5Ep5GOP/Q+cYp/75j0VRoGRG9EeZ0SKzdARIFZ4+QJEeX1naI+33kYuj7NeyHH8pmurSSEdvU9zUozloFhKsKZHyMf1EFGxkGkk8wygV9/CGhDfwb6HNrJnXzL6UvNC+MHGXHEWC93KGiwC5FJcRoAk1HbyQfuO5FBG2rGuG30IHOJYman6ak8U1Na9wLhRT+3fnRNFFpDFBUhtXdNTyWOHuhQg2alnX1Rk2oH/3ar6Vk5ZQ5kjBUgnHhd2tSeXEiDRkcWSI51Ls1oUIDada43tJMtVP7zZwzl2+zepgjD2WP8zZSRLzEBd5L556F6GaU+3MC72blxYvAyt09PgiLYfLtt0KY5++o2waw6i6w0igmRR5u23kcvivaN8x2/aFyhTr1hbXPWeKTuovp3n1ynZkKawW+idqAD7nfwmiffRRvbqSxZia6yL62dMM6xLWyqvGUOAXIoLCZDWDZYiKesyi4Fq89MHsyqlPODLEhznvqHuWbZtRhQad0yA1NY1Xxc/lr1cd68uuZfw4lnU1ncsn5CrrVxfgDib7i02gauIe89mHFqORq4VIDVhW8Gi2YsKkGyF4yOSFeUEzsT8gfn7zz6RCEBH19xe7L6NZ3XSMeePthnQ9jYSn3k3RtlF431k9tfMM9axkV93ZHjuzzzBEQ3RqrzWG2kjF2MKxfX2dQhH4vIipChA2gY88+OHYbnlZli/9qDZp7ihNrJfX+Jk3soODLu+TI3/hgC5FBcSIDHxUHY2smsEjDEpJ3icTs86L5sESEPds9QvGq+6pmh5tXUt1CURJpWuu/dscvnrF5/V39tpt+shNCc/+gMp6gVI3csvbQdOCMMQFxzO4GWcqlxnHtul+FIhWFNdvfbtvwwr7o1fx9Y6Z0O21r74VoQy7nbflhUxWsn42sF4Lbz2s3X/hFDUuCOvLZvhjSEgUvi7L9vyn+SX/BqQ4D7VC5BbaSOXYHoHiMdwbVVyFj0hQqoEiL8uS3mbFRpT66BOI+25d2Pmt+lQzfbIgFtqI/v0JTVizYu6EP0QslW6fyxCvxRXFyDVKrJSgFj1bNeKuAp73xmQvbJN1ZVTfU0ZAVKu620JkJZ7q/XrckS9/qdgKgVI0i5a0/DOMcLhVPcGARKL/805iBum/5flD/VdOQOyTYCkZyFL5WzbnC98se5531zRMcaU/yV/XSFAjCkNxW9ysCpnb7IhHe4eBUP7aH42FQtmb6GN7Iczg9olkgYUBysi9624BsRpX06SFCEevMGvSJ84JHdZBifYPjUciKkhMhvjh/5Vc1ttZP8+OOwXUr7Tp6o1vMyAXIorCpCG8KPqMtLO6yVCsLbv/FpTTsM1ZUKwVq2TcT/dJQSrRoBsubd7PZdzUS9A2nKit2fBWh+CFdtH4LIzIMYspvij33Ha43D//n2xmVhms1jeg6Y6Z0YYtzjWStpEHC0bue193xbJQUzBiZ+/VR+WNJ7paAEyVyQz8jq0I/HQ3rfdUBvZFW3taNVgxfD7aDKUmhCsYW+Q2JqH4UfLlLhO9rbwlJ1pDRtc1m2sx+v6Z3FjbWTfviRyDYnrrU83jAC5FFcUIM60V/gjm6rOjeesESAF8bCfAGmoe4HiIvSWa8otQq+oa3ozqL0WodcIkLr6pjrG9ZuOnZcaAVJKULCfAFkzA2J/96kxs0yqTvUvv8J3XOdl+P/LjbTas+H4dc0JsvUbNA51chycmrrsdt8S5d+kAKkc5d0kQIZ79GbeEtkkc1W8nTaSL3LDc4ixxvGrmgHpKlLtW7t+6l8WKV/Dry0XcK9CO4M2Kxf+31ob2bcvUU64lDD/6/vELIcTVsUMyGFcVYC4TvTUaFJpeKtmQLxsJEN5Uoiw0W0UINV1LzHl5E6l4W24puQiucq6uiMqi+99WJWGd1m1egFSVd+pg3P2QnleOwV9bsoCJDezdLwAWYh18TC1iWMFSGxwxMnQM86GxER9TZ2zmZVaEmE4vxp3Hvf6yTFcJTeocloBcukZkGhGyMoY/xtrI5lSTS8+7LueZK0AaV0DMn0vNpObTus8hof+JtV0T9vTF8+haEI8TCHITWshb7CNXKYvWYbt+fuZjBmyxrTZebHEGpBLcWUBYhaxlG7MbxibWRnGFdtIT+8dgtVQ9xoiG+otOpHaazLGLDcsdLOo1G9EGNRljK+s3IgwjInN/T7xWdVGhD0bEe5AUYCsfEldQ4AEoYGjUB32nOlEP+1snDrCF1HkKG0SFlvLkclipPVrwn5rss7IcFRd2yxM/74UdguOladfJ1uLOkGOnYmP8aw2e923FGcUIHYgJxEqkhpVd357a20kXXhmB+u1XFmALG3ZhhDGns+czc0TJ9N7t2YvmmUWQX9vNOG8D8dQxtjzuNU2cpG+ZNog8XHaZ8rNerW4zlK2tAZhlo0SgYD9BQi8CzCk90vWfnUs1M//jnZmqvTk2KcWV377+nnI+rNlDYh9CeeE0b8vvXmSX2zmIfEwxW7nQmPqR9/0dB3TYlT/Ra7t99T3l6GuXphhxaL0xWdOKtrYfR3jybOx8U7dF1n1iiOnyxHEzhuM2e2+Jahf7FpymIbz9y/z9WwMwRrbVurIXqPWRutXo/t5EGbeJyEtMhYj6G6zu+E2kvxVYaBiDauEpZdFSasfnt1aASLVT/vMxqM48+6Nvif3nJl3HA+FkM225SZnyO0qvsya6QnbG28j+/UlerGprS+KXDEYXKez4Wkq8UpNm9XP/bpB6ZOCANmLMeVi5sUU7OR5MDmRUd4YEe6VogARvzdkFptHlMTHPjnjNbZ/P163OAPiOAh2c649Z7xsmuCwzHjO+nHRYnTGz/nteD9me5+dYJ+snY0Lb/e45mF0vXUDPmPml6rzlwvct2U5dSOOdTMg0wZl0TZYT+2gjJI5B9tJxTvcU61kXTiO7sP1hjfbRpLf3D/8yqwbMPN/M4pe0c8NzwqSUPgWndGhz6uZGdLqR+R+z4kIWp9tbCbidtvIzn2Jksm66eGzZLWnQSS4FgiQMxNZjzKPyrDr+HsF+wUAAIAjQYCcnWGB+2J0Z8XoBtwP2C8AAAAcCQIE4GRgvwAAAHAkCBCAk4H9AgAAwJEgQABOBvYLAAAAR4IAATgZ2C8AAAAcCQIE4GRgvwAAAHAkCBCAk4H9AgAAwJEgQABOBvYLAAAAR4IAATgZ2C8AAAAcCQIE4GRgvwAAAHAkCBCAk4H9AgAAwJGsFiAcHBwcHBwcHBwcHBxrjlUCBADuE+wXAAAAjgQBAnAysF8AAAA4EgQIwMnAfgEAAOBIECAAJwP7BQAAgCNBgACcDOwXAAAAjgQBAnAysF8AAAA4EgQIwMnAfgEAAOBIECAAJwP7BQAAgCNBgACcDOwXAAAAjgQBAnAysF+AGNo8yS9G6f3KU99fjPbK0700v0lldjvNSnQvzVP/cng93gu6F0Z87DfcT2WkEEaqnzvW6r2ijBS/m3/0f/HPus6Ifs2T0N7z06YXnZFqVSXnGsnOdL7N+x1DJboXphOPZtXlJVDymHaHAAE4GS32O3Z2+zllALeJ7oXpOmH64UVs/91lDpF3AnRvROAIWefoeCfTOlaBU7SmpOJ9mo91TuFalPkkuiZRsN65s/dz6/VpJY3ouhUCVZlefMi3q6k9bvSmbwElM/anjCzZZoKpD9De31oUiO6N8MpQ0msbw7NYMxAxipk9Wd/utoEAATgZ7QJky6gewB0wOATuez3veJSdnJjtKJlwzq9tYwsnaRAjK8VDrYMWOGFzCVXnLx/SqEi5bb6jCMsZ2kb+nqx3eiOVMEI8mCYXM+uQO+UG9yhXXtv9v6a4jM4ozJ8mn0V+0iElItufrS9kdC/mcqMDE9Ul7yJ040VvmTlax3kFiJKRDgsuCvf8JmgWIDuPtgBkuXY/kXjxbhMgoaMQFfM1juMF8EdRlcw76rnPtwuQdSycuvg3QgGi1TTDFUXJSjHola1kQz9p29t+QqtyVLxVgETuQ+pZ7/1s84yziBlB33DvnItIfr7mPajkPLAwtdVK8dEyq9giCNvvV+W9WwkC5Oh6AFyZFvtd02GhV+BuGEe3p0arp3URmwTI4v2izV/y18j75vojjrY68RCRTQLkgFHyVQJkGNnPXkuzAKmYwanpFBdOf8Psjfc89XMfF1nvRYAUfbc1s1GlmYVZ9LSUqYMZELsubC1bw6+uKxTLIECOrsdq9lkcBeejVYDcUocFrdBP5NBquSB8DJ2Q6ueGNSDDCLfojX5Oh+9EjyuEYtlBhZ0FyL3MgJh4nP/iszqvfy475ktERF6qnL//7O0au6QAsQuuU2Ut7+sghkRvdHUYVcQPutkQrFHs7StAqoRnbrYyE6a32+xCi4BMXsL8nOoHDi43Q4sAOboeq1mjyAFa7PeC8aZwJegnqvHWgaydAZlmDcfwC/0aGVFWRvrvn+rwnw1MjuWtCpDcbMLyfq0VIGNdVocyLcr+GZ/FqnQWxzUnvfqZnQFJiib/PEUn+b5nQGaneb6OpCMtHo0QD0aIh7y4rxaLefEa+bb59vWz6TphZ0BFvyGZS+U6qULbbXpOVwgP3V+AjA1Xv5pPIqKMtRqmo2d1FU8FqI2SYqkqxWM4tVg63zT9PZ9PjsYejFpU1K14vgLaptoL6uN+pZfZ644anGtctfdY28wZwXmi4qzyeeTuUUr01dRXq6ms8dzHZ5K5T1oFyO7O6w59xDhi9Wbelt9NZLAp2ZRbL7/MOYuOtYHsuar7twJ320/M92l1331IPzHMWjiNfY0AGR2Uf1/6ZfYl36GLOXiXFiCDwPpF/hHUvSbU8johWIk+R/dBNqstAmQxU+D9re59br/7v743Uvwen0EoPEvtO3iFEKzQ+Z2dUtfRTtb/7gWItdH/9f1CvId1ithmao1O84Lw+R0g+kzGqLEfGkTHmF3N/s32SS13a7TP0iBB6TpS39G9XNSpTWit50IC5NFIEav8GPOqFi/BMR3gzNw5uIpxTBUmlj1n8XzL1GK2AT3JL95LrrJu2fOVCM8xTtFPIx3jy9e77rAxpEY2a+9xy71peB7GpO9R1LGoqa/9zlPvxE42jFrAknoBcqHR8z36CLeMycEcXw5eGsVamxrKfJJf5jKHc/8i//BypevIKGqt7RVv0J32E/azzX331fuJ0ZFrSb8ZOjmLa/Qd5tqQlosJkHHhszRv5i0qQNbOgOzLtQTIUKZ4cNppylbSo89jXH9ULGRu2Bzq5/yxYg1IECqklflH/2e0fh3aX0Zg3HUWrDn0yrZfb9ZnhQAZhfM86FFvn5Po9uxV69dh4Egs+qFF6NM04CIq0t4uB70uJkCe+6lOQjxcLfX+ZQRIojEmjdJfRJWLx/NfTqXzRcsJ4whr65Y7X4nU9G552jc2MhPvLLff48S9qX0e098SwiQyjV6s7w6xjzBTLUCa41orBeEOfUR6MWntwt6ITaXKTNW3NmSnUSzfaz+R/76p77uv2U9MI5WhEFq1BmRedZrYf0InBan6ftlNAfVzP9Q3dNDaBMjWLE7DET3hFQVIeJER4VtXjt8+886gDc2JXeM8o2e/k83YVaxjzXNqs5fDsiJqu0notCZmswDRphcf4s8o1Q68WSGtfjhCcH5X/iZVkO431ndPTn8yNExNouAf/V+xDW4RIEGdruRvXUiAxF625elQN6NEdTaCwvmyucudLCV1dcudr8S2ePqwAccci33ucezeNGWHSN2jwLBr63srm3e9D+pnQBpoWVO1uY/Ina/ezgKbSob+pASE2/Yb6p7lXvuJmrpX9t3X6if0vFj8zbyZXnzIzIDE7+Pff/bxNh9zmJVcOimLS57DOi7PVgFSpiwOkr+8uABJ7QWRFvgV9rsQv7Uzx/X7n1Slbb1CAoPbSMu+FCDVIYC196cx/Gz63BUkyw/M33/2iV3bdaQ9OrMe00zEtr165kuz31mXjnf/iJMLCZCYY1BW4/bmlYzXe8kVzpeeQXd/V1u33PlK1Iez6Ofei/+OjRjFytvpHkfuTfXzCH6fKrelvu7LWxjxMZFmEKq4DQGypY/IlRF3gqtsapMAaah7/ubcaT9RU/fKvvuq/YR2Ql5yAsR3UOewrUXI1/xjIz5664mGdTBTdiKp5nKuuhHh3gIkdKJuVoCk1mZENqIs1sljGl1e4SOkxE90Y8Ro3Z1ZS/26ejZt2z4Rsf7nEkQSOBjjKMttm0KmhMZq8bUqXFx769vKbbA8A9IqpJdlX+K5Xl2AlK/hOAFSdX8vLECmFInq56JTrxvZ3OkeX1mANI2s6Vfz7etn62SIR8KyVnAxAdIywrSpj8iVEbbHapvaQYBs76PvtZ+oKXebALlsP1EWIOPL2ZZbSC7gOsxDnP5YphAP5h/93ySqPl0905zvoO2wE3pk/cwWARKvw8YsWMkFx6XUrokNDcWH5feHfmLd84yEjmZF0VyPUJzbtQW2iPzsQH09nfOIx4PXX8YEiNOmN4Vsp2dxcwLk8sLtQAFywfW2VxQgtaEFa6bx28/XFmbknn6tAKk5R/pF2xJasfUe7xOCVeNYbAk3IUXsWsr229Jx2ufZFAKwQ/hUfRkNNrVDCNb29niv/URNuZV99yH9RIUAMcP9rVmg6Y/Yjxm8/EWsUm10mNawbYR4SXxNS87x3WsgtU2AZNbeOM5jvDwvJe4wyxZbQBzbY6XhguZy9WuQlS3xI/P3n/0wEPEaKTJ9j6ozV2knEcVgE5PtH/L6jc+AjO+gb18/rx+tzzjba2dAlBwzz22ZRdgjBKvC7iN9kZKXC7u7ogDJOSl2o52xMbcvZIyfz74sti60XtZtvQBJT2NNfy84OmXHov469rk3Jn4/qh2LuvoenXv8vXGJGZCmznmHPqJagLTY1CYB0lD3AvfaT+TLNfV99yH9RJ0AiY5UR4sbBIiT8caGag17A4wbxQ3XdKkQhzg7CpBx8a2fDWj1DEjLqesFSKodu5mQxv8fPgZbzlP/MiUsiNqytxB53UW5Mw3bw/K2ChA/e51rE9dK1RqSCMEys5hce/tzdrjK33BnsRyB2c4OMyA1vuuVB0OuKkDcF6Q7MtSahrdqtN05X0sa3mLdNgiQeQo4lV5zHqVx6yCFndaNORZhg6u8Dne0ZfG9D81peMN3dL1jUVXfyYidPQ6eK6anIcolBEj7Bkcb+4jqGZAGm9ooQOr7twJ32084ddvSdx/ST+QFiFZyntGoyYTohWBpbf/3k3DuZSTUNZ4RbG8nbz8BMs4IffLa4y0JELs/RthO54X/cxnxGYxRFIR78bh1GYXH2ObKKVbD+o5hg0I8TOGDTesdh5k2NwRrlQCZ9txZXrMvyse9iLLXunsbTgkQd5ZAtK8TzdazMaOamfse/x0zCva22aPtAqRqkPB9CxDjNOy5odRuRCjEQ3wzq5bz5TbRqt2IcMsDimyUtbim2KZfOq7MlxuROXVq2GAsqItOXWPl88jdo6SDV7PBWM9GhDuxuwBJCdHM9zf3ES1hXLU2tVmAVNa9hrvtJ4zZ3Hcf0k8kBIizedhik8dBhCQdL2cG5NvXz9MsyOIe+OumptmER2+m7zICZEzt2Rqrvgyhm9dA+RuvHS9AZuGweK7PfWYT4Xm/FHcA7kl+iTqM7v4J0rPP6N/jV7JYr+TvaySctjzuUTRlVNLa/PvSO3YhzP/6fhFN0iJA3MQWsTTSUSfW7R/EY6KvvqwA8VPaTskoxsQPhRJ1bFBBK6O+vzj3t3K3dP06PY9o/+D0UeJjKjtW8KONAqRyHd3dCxC4e66V0g+OYZv92pflk/xinuSXqSNdHXIAd8v76id8AbLMchWNuHFGM2NOxpgFqxcfFvs7/P1nb/596SOzVcYKS89hVfL2ZkCma/ciF1yBOK5liB3q+8uKTE3Lvqe8g7UfjukK6cz1T8IqJRwcwTA4kfHRbDedqtdGtFr0n6NQSVVpmfFudhDHGZun/iWaXrh5EbqS3uaMkfISff24CV98bdr+AuTNvDni3tuk1NZo8ZzGzHiLUMGsGJ3vrxAPmfoP+7q4GQmLM1d+3UriICVAZpvI2UO4aa5fjBVbSl63T0eA7IG2nYMQD8kjmqrxQHLOw3XjkeHabLXfMfZXiIfFaBsUoJ+4YcIRQv1c0ba1jr+sS2sTq3ZBtvV6kl92dgh2CMFKbrRoppmqT6XZlRVtZ96B+rHu/nleuZJ19z21V4hT0JTNrFgF9SPSjubwsNZQoZbHVszc1PgM1i3E3rkNOzPSY2hkyU7HWYmg6rpf9QwClDRiEDit70P93Ff0+yoj7uLiyjmDbYPfM+cYk89cObIEAXJWInHm84jNNTOywLXBfqEa+gkAALgACJAzMyxcXUzJrlDwcF9gv9AE/QQAAOwMAgTgZGC/AAAAcCQIEICTgf0CAADAkSBAAE4G9gsAAABHggABOBnYLwAAABwJAgTgZGC/AAAAcCQIEICTgf0CAADAkSBAAE4G9gsAAABHggABOBnYLwAAABwJAgTgZGC/AAAAcCQIEICTgf0CAADAkawWIG/8x3/8d5f/Yb/8x3/8x3/8x3/8d+R/qwUIBwcHBwcHBwcHBwfHmmOVAAGA+wT7BQAAgCNBgACcDOwXAAAAjgQBAnAysF8AAAA4EgQIwMnAfgEAAOBIECAAJwP7BQAAgCNBgACcDOwXAAAAjgQBAnAysF8AAAA4EgQIwMnAfgEAAOBIECAAJwP7BQAAgCNBgACcjPdov0oK89S/GL22AN0bKX43vfqZKL8zv0lVV76S9d81xujn3jz1LzXf9MrUphedkaryRBmU7Ezn11lXXMGe9y0s3Pwlf932XBdNnBZKAAAPvklEQVR1EUb0a+sy1KiXRnzsTfyWKyO7zoi+5r4p8/efvVGRr+pe2HOsqugNthG4aZTsTCcebZvWamW7uyxKWptorxr2cMsgQABORp39Wmeqaz0yPXv0JaJ7IzphZMKBrUUraUQ3vEjXvAOUNJ1IveDsvah+aenh3uUcb63Nt6+fbZ27OodT98J0nTCuf6t7kb3niYKM8MpR0nOcdW9EVyEe9rxvsbK7rlKcldH9xjYy1il5vdr04oOp0R+xZzCXsf6e3WQbuVmU+SS6JsdW98J04jF8bkoaIR4u4Lw3iNqVuO1DSdt+tvbHiTOZ3r/fQxvKv0eG/jRpd5kzYg83DQIE4GTUC5A2B0n3IvuiDDps+yMjOhmMKGv9aqQQjY7Y8IJb8bKO1m3+0HSROuaroowUv4fX1Qv7ghWP5ql/aXBYUteWek6F0rwX8+LZDS/Smvu4+32bfzyJuF3Rg9O51qFTMlOnVgESuTeD6PKPuvreZhvZjr2u5sGQ4PDvd7vYs/ck8ty0Mn/JX+2zGgYTlNw2YGNZ9+xa8B3ysY/KiZCpHxsOqUwoJiKCIe60O/cz4eh3orf3t6k/eK/28H5AgLxHNr344b1zuwJEG/3cm0+DsyHEQxB+U/VSr3ZASteaqXf4peL5Ra8LI+il8uP1XzWiZ+YXuzbOs2t6ke5w36I0OJyR6967jfjOlu9gJT/POTlRARK5ny19+U22keMo9UfRvkyrZDihMaZsv1oZKfac2bq+ADHGGK1+ZPuo+TfOdblt2rlPY4ipNmUBMoY1TSjp2JE912+1NxF7uHkQIAAn4/YESDeJjk48ml79TL78yk5F8ocRhy/j6I5ORipEICtoMqwSIKWZnbWhTnoKWZ7vqzbqeynk6ZL3bSi7cI9cRyD22TpnQBlZeKZ5x6XBWQxGfmMzMy1O6621keNZJUCGgYTUfbIhWCsGEJIhdyVybSpth7b+e8wclRz4CgHi3LO8APGudehDls/C2kn5GWAP98ANC5D9Fgvddx0A9uWaAiQ9Gj2UPbyY/9H/xauh9WJd374CxNYvuJbh5flm3uwLPH6xkdhxu65jEVblr3FZIUCqnJ7FSGG0kIwgaHdE9r1vw0fPQx2HMlLrOUv349ICJO/U5H6fWldl27/6vpzta3F2b7GN5KtS40RuY5UAMfF1A4vPSg5BxWxoXDC456+5z7H6N4QBuvVd0ye1CJBpjUlagCzubXZWQQ/hbunEI/dmD2flhgXIxgWM76YOAPtSL0DaO9291oA4P1q8JMZMRs2sECDjSy7+Dow70n6MeOCwtL7sG0ZOc45T4hfm29fPpuuEvVZRn3lp9/s2OG3jAs7xWvw49LJjMWfOambM6BWEQKVF9D/6v6KdRN8fQXussbfMPb3BNlKqx7rFurlR/aV9rxUgxowDJ/H+ojVcaeFHNM+ExGYFdhYgmUGB9E/2FSD/vswDUXYhfFfs5+fEI/HEJvdkD2dlfwEyTCVPnYJ4DBczOQu2xo7VVbLREYCml7aNw3TLD19k0lsw9biI/SzWoXANbl168SE8TzI+UVsnK1M3Y8xs4Pp1jpmfvbSw7Jr61jw7uHvqBci1QrCsfWr9Ghyp6famOP/CaHluJD85tZ4REkoO/U1M9BQc2mUETmtssV68vLO/0nMog9KjY/84/M3afSkEau/7pvWrV0U5ZZUZr63Faa1tI9VlFjK2jelMq2LLowKkNHsSsccbbiOXIdH/RJzoLQJkEjqLthoL6xnvZ2o0XoezgaWRd/+6fAEyXWetAEnPvM2z0G7bK4dB7i1ARiEv1U/zl/zVtq3CTNAY/qTvpM+EkJ0FiG3oi9GnQI2OaeWch61VZMp+7exDWP44tT/N7o0OumsH0U4hVYe2a1i+4GyDf5JfIgJkNny/boFRKWkbfWzBWyBAaupb8+zgPXC4APFyzX/7+nkpfJ0XZHOKzMYOI+dIz7YS2afD+06Vsyt62884v0vVeXz5TgvXG4TWnGkrMtMwZBcbHSb3PozPbh7USDv8e923IlMYxL6pQfVzP2RZqy9zFBjRtKiBI1WqgBN6ol9NKED8xcyhPd56G7kM1xIgQ5mL1LoZn0QPyTMWsVTp9QquuM5uHeGLg4X9tAiQmG+TWHuRuc5yeFjYh7vZq7JrQHL+RsUszTnt4b7ZV4CUwilMxklICJVWARKbIs393alAZHQjXofaa0iHC4zTyOGUcfLl7IuKwcCiHaz33ar6Vjw7eB8cKUA68RhZWLgPlxIg0Wn6YjacTJrVogCxTkSNbSfL9cNCnHjmmNMTDS0ZM5IlZqAuct88dL/z/gp63vuhpcxlaN08qq3th8s2XTPCPfS9QjxMs0aBAAn+7ZZ5+23kMlxRgPgUM5LZ0fhR3C5m17QyT/KLN8PpRiXIeHu8igDxymny49zfOmWuWANivO8G9yErQM5qD/fNRWZA0qNK5SnP+bM1AmT9PgDGxBpUrA6115CvS5hPvDFrQ+4ltzC42vqWnh28F64vQJxN9xabwFXEvWczDi1nF9cKkJqwrWDR7EUFSLbC8VG6inKSserDrtzxRAA6OkJ7sfs2ntVJx1waYW3p7tvbyBxi4ztw6vvLFMrrt/txRj02GuqOls79rSc4oiFaldd6I23kMlxegKSupTyI6Zb7YfkuzTn1ud3H/Wd5EQHijebXtBMnA9Rei9Cj/3b/vmondPPO7eG+2VmAzFOLY/jEcu1C2eGYG+caAVL/G3eUIu3wxMqrvYZCXRJhUum6ewIlNxqz+Kz+nuefHbwX6gVIy1S7b7/GGKPnEb4hVjacYcw4VblN3yLT9RcLwZrq6tmf7xhU3Jvopl8Ndc6GbK2dVloRarnbfVtWxGglwzVt6VpEQ0da22xO1LizOC3hJ+P+AFKI4J3WdTYEN7sGJLhP9QLkVtrIZbiwAEm1z2hK2AhKLjNaqR9WXAQhd3WDpdEkFvNUxW4CxF2/VeNDXUyADHW1C70rbK1mMONd28N9s7sAGdH6dTnqaYxpExWXEyB29E7Y/Qa09/dKAVKu120JkJb7GH928F6oEiDJDrY1De8cNxsuQt8gQCKOQtZBzOwZUedIL53RtTMg2wRI2mkplbNtc75wcGbP++aKjjHO+i/56woBYtLD125d1jodlbM32TTAWjthW7Z9ND+bQhawW2kjl+GCWbCSC5jjIdOp8ubfO//2sj1NtlBwYoO2tFqA5IX2WL9vXz9Xt3E30mMSIKMIXCtAUn7N6hmQ924P983FBMiM2wBaQqQuFYKVLrclBKt8DdcIwaoRIFvC0raFtMFtUi9A4lna9suCtT4EK7aPwGVnQIwxxtlwKvqdhMMhHqKhDs11zoy6bXGslezML/IP+ywqy9j7vj3JL4sZ17q9PNp3iT5cgMwVycTZD+1IPLT3vTfURo6mTYBEslUNuI5otrycM+0LTschL4WZLmZUFs9wpzS83nWWH697DqcOrn2sSsPr3WPdO2toHAHi/r1YVezhltlVgKQeaBBfGO3ElZHi97qMExmKi9CTDVJFGlNmEXrFNaQ3W9prEXqNAKmrb82zg/dB1QxmIYHCfgJkzQyI/d0nkXs519HmSBe+E2R0kd6mfOG9a6lzzllZbaduaMnw/3+rqM+u9y1R/k0KkMqRz00CxA2HiWU5zFbxdtpIvsgNz6HhHLUCJPWudjMrjf8/PSmbmP2wBQV9w/z/a9ZyOvXZRYDoIQz998XgSNXmeYm9SBZ1a9wJ/c28eWsqbLlxATL7ajX9yD3Yw1nZdwZkejDOfhpeClzX0Z8efSaF7boRoFwaXi+DyXB+KUSYRi9Zh8prcBrn8nsfVqXhDWJVKwVIVX2rnh28B8oCJDfzdbwAWQwmiIepzR4rQDIOx2CPdpQv51AUzpe8V43ZfMZfjTuPezHm4+ZeuSxRpxUgl54B8R3SYuYlr3431EYypa4epW86S6UAEeIh+i6e94eYy5jCt0OlsnhugTMfcdrdqJDazJ0xAZIPrdJ2TyX1w3z7+tmGNo4iWTx4G3+G4lnr17goi4igRZ+QsJXyGhDnunxfyg3B0rEB4+Bkd2IP52X/ECzdb96IcC7K3SywIUVsZEO9xYLq2OaAOmXwiTo0bEQY1MUOBSSuKdyIUIiH+EaE1QKksr41zw7unqIAWdlxX0OABKGLo5Ae9sTpRG/U95fssahBZTan2LUGazky2Vy0fq1wKBLniqXc1jYL078vhZ3HY+Xp16kviIYyOP2A+BjP9LLXfUtxRgFiB5rC6xk3PMs5N7fYRtKFb8ho1EBZgMxJARa3zcnAlhx8jG3cNw3mqWEzvUVlFs/729fP8zt9XBe0OI3bB2uj1Y8pTMmfZcnNgMx2OieW8ZdJzTM74+apc5m5sPSxTlLY9bTSF85rBEhq75Rom0kPUt+VPZyYK6wBgRjZcCuAC5IfQKgYWXJfmFpPjn2Y5cry7evnIevPljUgesiMkhZG/7705kl+sZmHxMMwspkPjakfydfTdYwvouDlNjgL6vvLUFcvDLJiUfris6wjNI/GCvFQeJHaui+y/vmDMonfuAMhy0G/ne5bgvoFoCUnYjh//zJfz8YQrLFtpY7sNY4j0v08SDTvHZAWGaOT6O9HcMttJPmrojBYy9zW4v1N+P3lrKVaJEPIDow40QLR/W7cswyzD/WDqGEfPNrD0gmPCRAvbFxnUsJO1xvZq6ezaxyWERB6Gao2iCTxsZ82G9T61ajvL4Egc+0vJ8zG/VLcmQ71/cXORkWzky3F2z3aw5m5HwEypjTMdPzBrrsHkxMZ9TnFAfalKECcuOAyYxiAfRGlZuRG+wxHsAoveifn/bevn3eekdPDSKVfZvylPToa0RlJ57fj/Zj7o9kJ9sn2A0OI2S7psIfR9dYN+IyxL/Vl3S9x35bl1IVH1M2ATHH80TZYT+2gkZI5B9tJxTvcU61k3aJa3f+/vXu7QRAIogDal4VYgvVYjbEXi8EfiCzrAwjsSOacZP40QcOI1x3cyX2S3R+fIx8fuev41WtH6/PM93T6g8S8nazLpz2qldX6mE7d6TrnU7X/Ejv644pvj33c7nXPzd2EotrpfWQ0VlScbrdLvUJXrNz15/cw6dEf5/gaMV3Z+Hq9GUbqfwaG8jUdpx9yO04AOaI396MMI1aLRspgQ/oXAIgkgOytv8G9GANZkahhK/oXAIgkgEAy+hcAiCSAQDL6FwCIJIBAMvoXAIgkgEAy+hcAiCSAQDL6FwCIJIBAMvoXAIgkgEAy+hcAiCSAQDL6FwCIJIBAMvoXAIgkgEAy+hcAiLQ6gCillFJKKaXUmlocQAAAALYkgAAAAM0IIAAAQDMCCAAA0IwAAgAANCOAAAAAzQggAABAMwIIAADQzBNMaf8WAesGiAAAAABJRU5ErkJggg=="
    }
   },
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![%E5%9B%BE%E7%89%87.png](attachment:%E5%9B%BE%E7%89%87.png)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 为建模创建虚拟变量(dummies)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-11-14T02:46:08.512756Z",
     "start_time": "2019-11-14T02:46:08.497422Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0    a\n",
       "1    b\n",
       "2    c\n",
       "3    d\n",
       "4    a\n",
       "5    b\n",
       "6    c\n",
       "7    d\n",
       "dtype: category\n",
       "Categories (4, object): [a, b, c, d]"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cat_s = pd.Series(['a', 'b', 'c', 'd']*2, dtype='category')\n",
    "cat_s"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-11-14T02:46:17.412984Z",
     "start_time": "2019-11-14T02:46:17.383636Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>a</th>\n",
       "      <th>b</th>\n",
       "      <th>c</th>\n",
       "      <th>d</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   a  b  c  d\n",
       "0  1  0  0  0\n",
       "1  0  1  0  0\n",
       "2  0  0  1  0\n",
       "3  0  0  0  1\n",
       "4  1  0  0  0\n",
       "5  0  1  0  0\n",
       "6  0  0  1  0\n",
       "7  0  0  0  1"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.get_dummies(cat_s)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 链式编程技术"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "当对数据集进行一系列变换时，你可能发现创建的多个临时变量其实并没有在分析中用到。看下面的例子："
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "```python\n",
    "df = load_data()\n",
    "df2 = df[df['col2'] < 0]\n",
    "df2['col1_demeaned'] = df2['col1'] - df2['col1'].mean()\n",
    "result = df2.groupby('key').col1_demeaned.std()\n",
    "```"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "首先，DataFrame.assign方法是一个df[k] = v形式的函数式的列分配方法。它不是就地修改对象，而是返回新的修改过的DataFrame。因此，下面的语句是等价的"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "```python\n",
    "# Usual non-functional way\n",
    "df2 = df.copy()\n",
    "df2['k'] = v\n",
    "\n",
    "# Functional assign way\n",
    "df2 = df.assign(k=v)\n",
    "```"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "就地分配可能会比assign快，但是assign可以方便地进行链式编程：\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "```python\n",
    "result = (df2.assign(coll_demeaned=df2.col1 - df2.col1.mean())\n",
    "         .groupby('key')\n",
    "         .col1_demeaned.std())\n",
    "```"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "使用链式编程时要注意，你可能会需要涉及临时对象。在前面的例子中，我们不能使用load_data的结果，直到它被赋值给临时变量df。为了这么做，assign和许多其它pandas函数可以接收类似函数的参数，即可调用对象（callable）。为了展示可调用对象，看一个前面例子的片段"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "```python\n",
    "df = load_data()\n",
    "df = df[df['col2'] < 0 ]\n",
    "\n",
    "\n",
    "df = (load_data()\n",
    "     [lambda x: x['col2'] < 0])\n",
    "```"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "示例1：把整个过程写出单链模式"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "```python\n",
    "result = (load_data()\n",
    "         [lambda x:x['col2'] < 0]\n",
    "         .assign(col1_demeaned= lambda x:x.col1 - x.col1.mean())\n",
    "         .groupby('key')\n",
    "         .col1_demeaned\n",
    "         .std())\n",
    "\n",
    "```"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 管道模式"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "你可以用Python内置的pandas函数和方法，用带有可调用对象的链式编程做许多工作。但是，有时你需要使用自己的函数，或是第三方库的函数。这时就要用到管道方法。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "```python\n",
    "a = f(df, arg1=v1)\n",
    "b = g(a, v2, arg3=v3)\n",
    "c = h(b, arg4=v4)\n",
    "```"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "当使用接收、返回Series或DataFrame对象的函数式，你可以调用pipe将其重写：\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "```python\n",
    "result = (df.pipe(f, arg1=v1)\n",
    "         .pipe(g, v2, arg3=v3)\n",
    "         .pipe(h, arg4=v4))\n",
    "```"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "pipe的另一个有用的地方是提炼操作为可复用的函数。看一个从列减去分组方法的例子："
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "```python\n",
    "g = df.groupby(['key1', 'key2'])\n",
    "df['col1'] = df['col1'] - g.mean()\n",
    "```"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "假设你想转换多列，并修改分组的键。另外，你想用链式编程做这个转换。下面就是一个方法："
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-11-14T03:08:04.769376Z",
     "start_time": "2019-11-14T03:08:04.758919Z"
    }
   },
   "source": [
    "```python\n",
    "def group_demean(df, by, cols):\n",
    "    result = df.copy()\n",
    "    g = df.groupby(by)\n",
    "    for c in cols:\n",
    "        result[c] = df[c] - g[c].transform('mean')\n",
    "    return result\n",
    "```"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "然后可以写出如下："
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-11-14T03:09:06.567976Z",
     "start_time": "2019-11-14T03:09:06.540006Z"
    }
   },
   "source": [
    "```python\n",
    "result = (df[df.col1<0]\n",
    "         .pipe(group_demean, ['key1', 'key2'], ['col1']))\n",
    "\n",
    "```"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.3"
  },
  "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": false
  },
  "varInspector": {
   "cols": {
    "lenName": 16,
    "lenType": 16,
    "lenVar": 40
   },
   "kernels_config": {
    "python": {
     "delete_cmd_postfix": "",
     "delete_cmd_prefix": "del ",
     "library": "var_list.py",
     "varRefreshCmd": "print(var_dic_list())"
    },
    "r": {
     "delete_cmd_postfix": ") ",
     "delete_cmd_prefix": "rm(",
     "library": "var_list.r",
     "varRefreshCmd": "cat(var_dic_list()) "
    }
   },
   "types_to_exclude": [
    "module",
    "function",
    "builtin_function_or_method",
    "instance",
    "_Feature"
   ],
   "window_display": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}