16-Python Pandas聚合函数

Python Pandas聚合函数

窗口函数可以与聚合函数一起使用，聚合函数指的是对一组数据求总和、最大值、最小值以及平均值的操作。

应用聚合函数

首先让我们创建一个 DataFrame 对象，然后对聚合函数进行应用。

import pandas as pd
import numpy as np
df = pd.DataFrame(np.arange(20).reshape(5,4),index = pd.date_range('12/14/2020', periods=5),columns = ['A', 'B', 'C', 'D'])
print (df)
#窗口大小为3，min_periods 最小观测值为1
r = df.rolling(window=3,min_periods=1)
print(r)

输出结果：

             A   B   C   D
2020-12-14   0   1   2   3
2020-12-15   4   5   6   7
2020-12-16   8   9  10  11
2020-12-17  12  13  14  15
2020-12-18  16  17  18  19
Rolling [window=3,min_periods=1,center=False,axis=0,method=single]

1) 对整体聚合

您可以把一个聚合函数传递给 DataFrame，示例如下：

import pandas as pd
import numpy as np
df = pd.DataFrame(np.arange(20).reshape(5,4),index = pd.date_range('12/14/2020', periods=5),columns = ['A', 'B', 'C', 'D'])
print (df)
#窗口大小为3，min_periods 最小观测值为1
r = df.rolling(window=3,min_periods=1)
print(r.aggregate(np.sum))
# 以下方式也可以
# print(r.sum())

输出结果：

             A   B   C   D
2020-12-14   0   1   2   3
2020-12-15   4   5   6   7
2020-12-16   8   9  10  11
2020-12-17  12  13  14  15
2020-12-18  16  17  18  19
               A     B     C     D
2020-12-14   0.0   1.0   2.0   3.0
2020-12-15   4.0   6.0   8.0  10.0
2020-12-16  12.0  15.0  18.0  21.0
2020-12-17  24.0  27.0  30.0  33.0
2020-12-18  36.0  39.0  42.0  45.0

2) 对任意某一列聚合

import pandas as pd
import numpy as np
df = pd.DataFrame(np.arange(20).reshape(5,4),index = pd.date_range('12/14/2020', periods=5),columns = ['A', 'B', 'C', 'D'])
print (df)
#窗口大小为3，min_periods 最小观测值为1
r = df.rolling(window=3,min_periods=1)
print(r['B'].aggregate(np.sum))

输出结果：

             A   B   C   D
2020-12-14   0   1   2   3
2020-12-15   4   5   6   7
2020-12-16   8   9  10  11
2020-12-17  12  13  14  15
2020-12-18  16  17  18  19
2020-12-14     1.0
2020-12-15     6.0
2020-12-16    15.0
2020-12-17    27.0
2020-12-18    39.0
Freq: D, Name: B, dtype: float64

3) 对多列数据聚合

import pandas as pd
import numpy as np
df = pd.DataFrame(np.arange(20).reshape(5,4),index = pd.date_range('12/14/2020', periods=5),columns = ['A', 'B', 'C', 'D'])
print (df)
#窗口大小为3，min_periods 最小观测值为1
r = df.rolling(window=3,min_periods=1)
print(r['B','C'].aggregate(np.sum))

输出结果：

             A   B   C   D
2020-12-14   0   1   2   3
2020-12-15   4   5   6   7
2020-12-16   8   9  10  11
2020-12-17  12  13  14  15
2020-12-18  16  17  18  19
               B     C
2020-12-14   1.0   2.0
2020-12-15   6.0   8.0
2020-12-16  15.0  18.0
2020-12-17  27.0  30.0
2020-12-18  39.0  42.0

4) 对单列应用多个函数

import pandas as pd
import numpy as np
df = pd.DataFrame(np.arange(20).reshape(5,4),index = pd.date_range('12/14/2020', periods=5),columns = ['A', 'B', 'C', 'D'])
print (df)
#窗口大小为3，min_periods 最小观测值为1
r = df.rolling(window=3,min_periods=1)
print(r['B'].aggregate([np.sum,np.mean]))

输出结果：

             A   B   C   D
2020-12-14   0   1   2   3
2020-12-15   4   5   6   7
2020-12-16   8   9  10  11
2020-12-17  12  13  14  15
2020-12-18  16  17  18  19
             sum  mean
2020-12-14   1.0   1.0
2020-12-15   6.0   3.0
2020-12-16  15.0   5.0
2020-12-17  27.0   9.0
2020-12-18  39.0  13.0

5) 对不同列应用多个函数

import pandas as pd
import numpy as np
df = pd.DataFrame(np.arange(20).reshape(5,4),index = pd.date_range('12/14/2020', periods=5),columns = ['A', 'B', 'C', 'D'])
print (df)
#窗口大小为3，min_periods 最小观测值为1
r = df.rolling(window=3,min_periods=1)
print(r['B','C'].aggregate([np.sum,np.mean]))

输出结果：

             A   B   C   D
2020-12-14   0   1   2   3
2020-12-15   4   5   6   7
2020-12-16   8   9  10  11
2020-12-17  12  13  14  15
2020-12-18  16  17  18  19
               B           C      
             sum  mean   sum  mean
2020-12-14   1.0   1.0   2.0   2.0
2020-12-15   6.0   3.0   8.0   4.0
2020-12-16  15.0   5.0  18.0   6.0
2020-12-17  27.0   9.0  30.0  10.0
2020-12-18  39.0  13.0  42.0  14.0

6) 对不同列应用不同函数

import pandas as pd
import numpy as np
df = pd.DataFrame(np.arange(20).reshape(5,4),index = pd.date_range('12/14/2020', periods=5),columns = ['A', 'B', 'C', 'D'])
print (df)
#窗口大小为3，min_periods 最小观测值为1
r = df.rolling(window=3,min_periods=1)
print(r.aggregate({"B":np.sum,"C":np.mean}))

输出结果：

             A   B   C   D
2020-12-14   0   1   2   3
2020-12-15   4   5   6   7
2020-12-16   8   9  10  11
2020-12-17  12  13  14  15
2020-12-18  16  17  18  19
               B     C
2020-12-14   1.0   2.0
2020-12-15   6.0   4.0
2020-12-16  15.0   6.0
2020-12-17  27.0  10.0
2020-12-18  39.0  14.0