7-pandas分组计算

2019独角兽企业重金招聘Python工程师标准>>>

#encoding:utf8


import numpy as np
import pandas as pd


'''
分组计算：
    拆分：根据什么进行分组
    应用：每个分组进行怎样的计算
    合并：每个分组的计算结果合并起来
'''

df = pd.DataFrame(
    {
        'key1':['a','a','b','b','a'],
        'key2':['one','two','one','two','one'],
        'data1':np.random.randint(1,10,5),
        'data2':np.random.randint(1,10,5),
    }
)
print(df)
'''
   data1  data2 key1 key2
0      3      9    a  one
1      3      5    a  two
2      7      3    b  one
3      1      1    b  two
4      2      1    a  one
'''
print(df['data1'])
'''
0    7
1    7
2    4
3    1
4    8
'''
#计算data1列按照key1字段聚合求平均值
print(df['data1'].groupby(df['key1']).mean())
'''
   data1  data2 key1 key2
0      1      3    a  one
1      4      4    a  two
2      3      9    b  one
3      7      3    b  two
4      1      3    a  one

key1
a    2
b    5
'''
#除了可以在数据内按照聚合，也可以自定义聚合
#这里列表中元素表示data1列的元素位置聚合
#data1列下的第1,3,4个元素聚合，第2,5个元素聚合，然后求平均值
key = [1,2,1,1,2]
print(df)
print(df['data1'].groupby(key).mean())
'''
   data1  data2 key1 key2
0      9      9    a  one
1      5      2    a  two
2      4      8    b  one
3      2      4    b  two
4      1      3    a  one

1    5
2    3
'''

#分组也可以是多级列表
#在groupby中按照元素索引顺序进行排组的依据先后
print(df)
print(df['data1'].groupby([df['key1'],df['key2']]).sum())
print(df['data1'].groupby([df['key2'],df['key1']]).sum())
'''
   data1  data2 key1 key2
0      9      8    a  one
1      3      4    a  two
2      7      8    b  one
3      4      1    b  two
4      3      7    a  one
a[9,3,3]---->one[9,3]....

key1  key2
a     one     12
      two      3
b     one      7
      two      4

one[9,7,3]---->a[9,3]....
key2  key1
one   a       12
      b        7
two   a        3
      b        4
'''

#查看分组的个数
print(df['data1'].groupby([df['key1'],df['key2']]).size())
'''
key1  key2
a     one     2
      two     1
b     one     1
      two     1
'''

#按照key1进行分组
#生成的是一个DataFrame
print(df.groupby('key1').sum())
'''
      data1  data2
key1
a        16     12
b        10      6
'''
#对分组之后可进行索引的选取
print(df.groupby('key1').sum()['data1'])
'''
key1
a    14
b    15
'''
#当然也可以进行多级分组，然后转换为DataFrame
mean = df.groupby(['key1','key2']).sum()
print(mean)
print(mean.unstack())
'''
           data1  data2
key1 key2
a    one      17      6
     two       1      3
b    one       7      3
     two       1      6

     data1     data2
key2   one two   one two
key1
a       17   1     6   3
b        7   1     3   6
'''
#groupby支持迭代
for name,group in df.groupby('key1'):
    print(name)
    print(group)
'''
a
   data1  data2 key1 key2
0      9      8    a  one
1      8      8    a  two
4      7      7    a  one

b
   data1  data2 key1 key2
2      6      1    b  one
3      5      8    b  two
'''
#也可以对groupby转换为字典
print(dict(list(df.groupby('key1'))))
print(dict(list(df.groupby('key1')))['a'])
print(dict(list(df.groupby('key1')))['b'])
'''
{'a':    data1  data2 key1 key2
0      9      4    a  one
1      4      3    a  two
4      5      6    a  one, 'b':    data1  data2 key1 key2
2      3      7    b  one
3      2      8    b  two}

   data1  data2 key1 key2
0      9      4    a  one
1      4      3    a  two
4      5      6    a  one

   data1  data2 key1 key2
2      3      7    b  one
3      2      8    b  two
'''
#按照列类型进行分组
print(df.groupby(df.dtypes,axis=1).sum())
'''
0      4   aone
1     15   atwo
2     10   bone
3      8   btwo
4     18   aone
'''

#以上都是按照列表进行分组
#下面用其他分组形式来进行分组
#通过字典进行分组
df = pd.DataFrame(
    np.random.randint(1,10,(5,5)),
    columns=list('abcde'),
    index=['Alice','Bob','Candy','Dark','Emily']
)
#看一下处理非数字
df.ix[1,1:3] = np.NaN
print(df)
'''
       a    b    c  d  e
Alice  3  3.0  7.0  7  9
Bob    4  NaN  NaN  3  4
Candy  9  5.0  1.0  4  1
Dark   6  3.0  9.0  9  9
Emily  8  4.0  2.0  3  6
'''
mapping = {
    'a':'red',
    'b':'red',
    'c':'blue',
    'd':'orange',
    'e':'blue'
}
grouped = df.groupby(mapping,axis=1)
print(grouped.sum())
'''
       a    b    c  d  e
Alice  4  5.0  2.0  2  2
Bob    3  NaN  NaN  8  1
Candy  6  7.0  6.0  9  7
Dark   7  8.0  1.0  4  3
Emily  8  6.0  8.0  3  3

       blue  orange   red
Alice   4.0     2.0   9.0
Bob     1.0     8.0   3.0
Candy  13.0     9.0  13.0
Dark    4.0     4.0  15.0
Emily  11.0     3.0  14.0
可以看对Nan和数字分组计算是按照Nan=0来处理的
'''
print(grouped.size())
print(grouped.count())
'''
blue      2
orange    1
red       2

       blue  orange  red
Alice     2       1    2
Bob       1       1    1
Candy     2       1    2
Dark      2       1    2
Emily     2       1    2
Nan是没有统计个数的
'''


#通过函数进行分组
df = pd.DataFrame(
    np.random.randint(1,10,(5,5)),
    columns=list('abcde'),
    index=['Alice','Bob','Candy','Dark','Emily']
)
#按照行索引
def _group_by(idx):
    print(idx)
    return idx
print(df.groupby(_group_by).size())
print(df.groupby(_group_by).count())
'''
Alice
Bob
Candy
Dark
Emily
       a  b  c  d  e
Alice  1  1  1  1  1
Bob    1  1  1  1  1
Candy  1  1  1  1  1
Dark   1  1  1  1  1
Emily  1  1  1  1  1
按照行来进行分组的
'''
#按照行索引长度
def _group_by2(idx):
    print(idx)
    return len(idx)
print(df.groupby(_group_by2).size())
print(df.groupby(_group_by2).count())
'''
Alice
Bob
Candy
Dark
Emily

   a  b  c  d  e
3  1  1  1  1  1
4  1  1  1  1  1
5  3  3  3  3  3
'''


#通过索引级别进行分组
columns = pd.MultiIndex.from_arrays(
    [
        ['China','USA','China','USA','China'],
        ['A','A','B','C','B']
    ],
    names=['country','index']
)
df = pd.DataFrame(np.random.randint(1,10,(5,5)),columns=columns)
print(df)
'''
country China USA China USA China
index       A   A     B   C     B
0           9   5     4   5     2
1           8   5     4   6     9
2           2   7     3   1     3
3           2   1     6   5     5
4           5   3     6   9     4
'''
print(df.groupby(level='country',axis=1).sum())
'''
country  China  USA
0           17   12
1           19   12
2           19    8
3           21   13
4           18    9
'''
print(df.groupby(level='index',axis=1).sum())
'''
index   A   B  C
0       7   7  6
1      11  12  3
2      12  11  9
3      14  11  4
4      17  15  8

'''

 

转载于:https://my.oschina.net/eddylinux/blog/1529788