打卡40天

pandas协方差和相关系数：

协方差和相关系数表示两个数据集之间存在的相关关系，在pandas中使用cov方法求协方差，使用corr方法来求相关系数，相关系数=协方差/标准差

#转换成矩阵
matrix_newdata = np.array(new_data)
print(matrix_newdata)

'''[[1.11397015e+03 4.26950000e+03 2.88167043e+03 1.92434336e+04
  3.15348674e+01]
 [1.26969300e+03 4.30440000e+03 3.42342447e+03 2.10167620e+04
  3.38999825e+01]
 [1.32771597e+03 4.33480000e+03 4.17657786e+03 2.55242657e+04
  4.06121790e+01]
 [1.60106904e+03 4.36370000e+03 5.01189343e+03 3.18399677e+04
  5.06839994e+01]
 [1.93002764e+03 4.38900000e+03 5.86391531e+03 3.72373131e+04
  5.90975434e+01]]'''

#转置
df_matrix_newdata = pd.DataFrame(matrix_newdata.T)
print(df_matrix_newdata)

'''              0             1             2             3             4
0   1113.970150   1269.693003   1327.715967   1601.069038   1930.027639
1   4269.500000   4304.400000   4334.800000   4363.700000   4389.000000
2   2881.670431   3423.424472   4176.577856   5011.893427   5863.915310
3  19243.433578  21016.761992  25524.265651  31839.967740  37237.313050
4     31.534867     33.899983     40.612179     50.683999     59.097543'''

print(df_matrix_newdata[0].cov(df_matrix_newdata[1]))
print(df_matrix_newdata[0].corr(df_matrix_newdata[1]))

'''协方差：67146730.45383753'''
'''相关系数：0.9996476727699803'''

除了pandas中的cov和corr方法，可以使用numpy直接求协方差矩阵和相关系数矩阵

np.cov(matrix_newdata)

'''
协方差矩阵
array([[6.15981984e+07, 6.71467305e+07, 8.20479822e+07, 1.02863697e+08,
        1.20457241e+08],
       [6.71467305e+07, 7.32466577e+07, 8.95791295e+07, 1.12382902e+08,
        1.31668519e+08],
       [8.20479822e+07, 8.95791295e+07, 1.09699667e+08, 1.37783328e+08,
        1.61545852e+08],
       [1.02863697e+08, 1.12382902e+08, 1.37783328e+08, 1.73247291e+08,
        2.03265037e+08],
       [1.20457241e+08, 1.31668519e+08, 1.61545852e+08, 2.03265037e+08,
        2.38588380e+08]])
'''

np.corrcoef(matrix_newdata)

'''
相关系数矩阵
array([[1.        , 0.99964767, 0.99811652, 0.99573758, 0.99362901],
       [0.99964767, 1.        , 0.99933311, 0.99763829, 0.99600931],
       [0.99811652, 0.99933311, 1.        , 0.99944935, 0.998547  ],
       [0.99573758, 0.99763829, 0.99944935, 1.        , 0.99978085],
       [0.99362901, 0.99600931, 0.998547  , 0.99978085, 1.        ]])
'''

聚合统计：

print(new_data.aggregate(np.mean))

'''
社会消费品零售总额(亿元)     1448.495159
常驻人口              4332.280000
地区生产总值            4271.496299
人均可支配收入(元)       26972.348402
居民消费价格指数            43.165714
dtype: float64
'''

pandas中的aggregate方法不仅可以对整个数据帧进行聚合，对单个列，多个列，单列多种聚合运算，多列多种聚合运算也同样适用

print(new_data.aggregate(np.mean))

'''
社会消费品零售总额(亿元)     1448.495159
常驻人口              4332.280000
地区生产总值            4271.496299
人均可支配收入(元)       26972.348402
居民消费价格指数            43.165714
dtype: float64
'''