ch01/特征探索性分析.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import scipy.stats as ss\n",
    "import seaborn as sns"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 假设检验"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([ 0.75295476,  1.23743465,  1.22311872,  1.03186398,  0.02225342,\n",
       "       -1.5607541 ,  0.01890839, -1.56292672, -1.12690071, -0.13936368,\n",
       "        0.51524727,  1.21290744,  1.52412826,  0.67158692, -0.12853659,\n",
       "       -0.07320661,  0.68413995, -1.07913583,  0.34829248, -0.25102207])"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 20个符合正态分布的值\n",
    "norm_dist = ss.norm.rvs(size=20)\n",
    "norm_dist"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "NormaltestResult(statistic=1.4377483819665124, pvalue=0.4873005546205281)"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 检验是否符合正态分布\n",
    "ss.normaltest(norm_dist)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Ttest_indResult(statistic=-1.5743362836198633, pvalue=0.1266426208728476)"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# t分布检验\n",
    "ss.ttest_ind(ss.norm.rvs(size=10), ss.norm.rvs(size=20))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "检验统计量和P值。P值大于0.05, 也就是可以说假设成立，两分布均值没有差别"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(126.08080808080808, 2.9521414005078985e-29, 1, array([[55., 55.],\n",
       "        [45., 45.]]))"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 卡方检验\n",
    "ss.chi2_contingency([[15, 95], [85, 5]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "F_onewayResult(statistic=17.619417475728156, pvalue=0.0002687153079821641)"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# f检验\n",
    "ss.f_oneway([49, 50, 39, 40, 43],[28, 32, 30, 26, 34],[38, 40, 45, 42, 48])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "from statsmodels.graphics.api import qqplot\n",
    "from matplotlib import pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# 图检验是否符合正态分布\n",
    "plt.show(qqplot(ss.norm.rvs(size=100)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 相关系数"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.9333729600465923"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import pandas as pd\n",
    "s1 = pd.Series([0.1, 0.2, 1.1, 2.4, 1.3, 0.3, 0.5])\n",
    "s2 = pd.Series([0.5, 0.4, 1.2, 2.5, 1.1, 0.7, 0.1])\n",
    "# 对Seres求相关系数\n",
    "s1.corr(s2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.7142857142857144"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 斯皮尔曼相关系数\n",
    "s1.corr(s2, method='spearman')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "      <th>1</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.933373</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.933373</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          0         1\n",
       "0  1.000000  0.933373\n",
       "1  0.933373  1.000000"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 对df求相关系数，因为对列处理，所以要转置\n",
    "df=pd.DataFrame(np.array([s1,s2]).T)\n",
    "df.corr()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "      <th>1</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.714286</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.714286</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          0         1\n",
       "0  1.000000  0.714286\n",
       "1  0.714286  1.000000"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.corr(method='spearman')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 对于回归问题\n",
    "x=np.arange(10).astype(np.float).reshape((10, 1))\n",
    "y=x*3+np.random.random((10, 1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0.],\n",
       "       [1.],\n",
       "       [2.],\n",
       "       [3.],\n",
       "       [4.],\n",
       "       [5.],\n",
       "       [6.],\n",
       "       [7.],\n",
       "       [8.],\n",
       "       [9.]])"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[ 0.38241749],\n",
       "       [ 3.02216578],\n",
       "       [ 6.03322823],\n",
       "       [ 9.66856137],\n",
       "       [12.9455134 ],\n",
       "       [15.8189476 ],\n",
       "       [18.94890963],\n",
       "       [21.33221138],\n",
       "       [24.67287576],\n",
       "       [27.70559378]])"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[ 0.28923468],\n",
       "       [ 3.34785863],\n",
       "       [ 6.40648258],\n",
       "       [ 9.46510652],\n",
       "       [12.52373047],\n",
       "       [15.58235441],\n",
       "       [18.64097836],\n",
       "       [21.69960231],\n",
       "       [24.75822625],\n",
       "       [27.8168502 ]])"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.linear_model import LinearRegression\n",
    "# 线性回归器\n",
    "reg = LinearRegression()\n",
    "# 对数据拟合\n",
    "res=reg.fit(x,y)\n",
    "# 产生预测值\n",
    "y_pred=reg.predict(x)\n",
    "y_pred"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[3.05862395]])"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 参数\n",
    "reg.coef_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0.28923468])"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 截距\n",
    "reg.intercept_"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### PCA变换"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "data=np.array([np.array([2.5, 0.5, 2.2, 1.9, 3.1, 2.3, 2,   1,  1.5, 1.1]), \n",
    "               np.array([2.4, 0.7, 2.9, 2.2,  3,  2.7, 1.6, 1.1, 1.6, 0.9])]).T"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "PCA(copy=True, iterated_power='auto', n_components=1, random_state=None,\n",
       "  svd_solver='auto', tol=0.0, whiten=False)"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.decomposition import PCA\n",
    "lower_dim = PCA(n_components=1)\n",
    "lower_dim.fit(data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0.96318131])"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "lower_dim.explained_variance_ratio_"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "降维后留下0.96信息量"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[-0.82797019],\n",
       "       [ 1.77758033],\n",
       "       [-0.99219749],\n",
       "       [-0.27421042],\n",
       "       [-1.67580142],\n",
       "       [-0.9129491 ],\n",
       "       [ 0.09910944],\n",
       "       [ 1.14457216],\n",
       "       [ 0.43804614],\n",
       "       [ 1.22382056]])"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 降维之后的数据\n",
    "lower_dim.fit_transform(data)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### sk-learn中PCA算法用到是奇异值分解"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 自己实现一个PCA方法"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(array([[-0.82797019],\n",
      "       [ 1.77758033],\n",
      "       [-0.99219749],\n",
      "       [-0.27421042],\n",
      "       [-1.67580142],\n",
      "       [-0.9129491 ],\n",
      "       [ 0.09910944],\n",
      "       [ 1.14457216],\n",
      "       [ 0.43804614],\n",
      "       [ 1.22382056]]), array([0.0490834 +0.j, 1.28402771+0.j]))\n"
     ]
    }
   ],
   "source": [
    "def myPCA(data, n_components=1000000):  # 默认取所有维度\n",
    "    # data中针对列计算均值\n",
    "    mean_vals = np.mean(data, axis=0)\n",
    "    # 每列分别减去均值\n",
    "    mid = data - mean_vals\n",
    "    # 求协方差 np.cov， rowvar=False针对列\n",
    "    cov_mat = np.cov(mid, rowvar=False)\n",
    "    from scipy import linalg # 线性计算\n",
    "    # 求协方差矩阵的特征值 和 特征向量\n",
    "    eig_vals, eig_vects = linalg.eig(np.mat(cov_mat))\n",
    "    # np.argsort得到排序后的索引\n",
    "    eig_val_index = np.argsort(eig_vals)\n",
    "    # 取最大索引\n",
    "    eig_val_index = eig_val_index[:-(n_components+1):-1]\n",
    "    # 取特征向量\n",
    "    eig_vects = eig_vects[:, eig_val_index]\n",
    "    # 转换后的特征\n",
    "    low_dim_mat = np.dot(mid, eig_vects)\n",
    "    return low_dim_mat, eig_vals\n",
    "print(myPCA(data, n_components=1))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "结果一致，哦耶！"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 基于HR.csv数据做主成分分析"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([9.98565340e-01, 8.69246970e-04, 4.73865973e-04, 4.96932182e-05,\n",
       "       2.43172315e-05, 9.29496619e-06, 8.24128218e-06])"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# HR.bak.csv是清洗过后的数据\n",
    "df = pd.read_csv('HR.bak.csv')\n",
    "my_pca = PCA(n_components=7)\n",
    "lower_mat = my_pca.fit_transform(df.drop(labels=['salary','department','left'], axis=1))\n",
    "my_pca.explained_variance_ratio_"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "得出相关系数，比较后，只留下salary一列即可"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7f333cba1c88>"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "sns.heatmap(pd.DataFrame(lower_mat).corr(), vmin=1, vmax=1, cmap=sns.color_palette('RdBu', n_colors=128))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 交叉分析"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Ttest_indResult(statistic=-1.0601649378624074, pvalue=0.2891069046174478)\n"
     ]
    }
   ],
   "source": [
    "# indices获得索引\n",
    "df_indices = df.groupby(by='department').indices\n",
    "sales_values = df['left'].iloc[df_indices['sales']].values\n",
    "technical_values = df['left'].iloc[df_indices['technical']].values\n",
    "# t分布校验\n",
    "print(ss.ttest_ind(sales_values, technical_values))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "df_keys = list(df_indices.keys())\n",
    "# 初始化一个为0的矩阵\n",
    "df_t_mat = np.zeros([len(df_keys), len(df_keys)])\n",
    "for i in range(len(df_keys)):\n",
    "    for j in range(len(df_keys)):\n",
    "        p_value = ss.ttest_ind(df['left'].iloc[df_indices[df_keys[i]]].values, \n",
    "                               df['left'].iloc[df_indices[df_keys[j]]].values)[1]\n",
    "        # 这样做图颜色对比更明显\n",
    "        if p_value < 0.05:\n",
    "            df_t_mat[i][j]=-1\n",
    "        else:\n",
    "            df_t_mat[i][j] = p_value\n",
    "sns.heatmap(df_t_mat, xticklabels=df_keys, yticklabels=df_keys)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "所有黑色部分可以认为部门之间离职率是有显著差异的，例如IT部门和technical部门"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Work_accident</th>\n",
       "      <th>0</th>\n",
       "      <th>1</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>promotion_last_5years</th>\n",
       "      <th>salary</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th rowspan=\"3\" valign=\"top\">0</th>\n",
       "      <th>high</th>\n",
       "      <td>0.082996</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>low</th>\n",
       "      <td>0.331728</td>\n",
       "      <td>0.090020</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>medium</th>\n",
       "      <td>0.230683</td>\n",
       "      <td>0.081655</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"3\" valign=\"top\">1</th>\n",
       "      <th>high</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>low</th>\n",
       "      <td>0.229167</td>\n",
       "      <td>0.166667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>medium</th>\n",
       "      <td>0.028986</td>\n",
       "      <td>0.023256</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "Work_accident                        0         1\n",
       "promotion_last_5years salary                    \n",
       "0                     high    0.082996  0.000000\n",
       "                      low     0.331728  0.090020\n",
       "                      medium  0.230683  0.081655\n",
       "1                     high    0.000000  0.000000\n",
       "                      low     0.229167  0.166667\n",
       "                      medium  0.028986  0.023256"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 透视表是另一种交叉分析方法\n",
    "piv_tb = pd.pivot_table(df, values='left', index=['promotion_last_5years', 'salary'],\n",
    "                       columns=['Work_accident'], aggfunc=np.mean)\n",
    "piv_tb"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "可以得出0.331728的值比较高，过去5年没有晋升，工资比较低，没有事故，这样的人离职率高。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "sns.heatmap(piv_tb, vmin=0, vmax=1, cmap=sns.color_palette('Reds', n_colors=256))\n",
    "sns.set_context(font_scale=1.5)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 分组分析"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/f/anaconda3/lib/python3.7/site-packages/scipy/stats/stats.py:1713: FutureWarning: Using a non-tuple sequence for multidimensional indexing is deprecated; use `arr[tuple(seq)]` instead of `arr[seq]`. In the future this will be interpreted as an array index, `arr[np.array(seq)]`, which will result either in an error or a different result.\n",
      "  return np.add.reduce(sorted[indexer] * weights, axis=axis) / sumval\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "sns.barplot(x='salary', y='left', hue='department', data=df)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "可以看出图中两个拐点，可以以此进行分析"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 相关分析(除相关系数外，熵也是一个很重要的因素)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}