2018.03.29 python-pandas transform/apply 的使用

 1 #一般化的groupby方法:apply
 2 df = pd.DataFrame({'data1':np.random.rand(5),
 3                    'data2':np.random.rand(5),
 4                    'key1':list('aabba'),
 5                    'key2':['one','two','one','two','one']})
 6 print(df)
 7 #print(df.groupby('key1').apply(lambda x:x.describe()))
 8 #apply直接运算其中的函数
 9 #这里是匿名函数，直接描述分组后的统计量
10 
11 def f_df1(d,n):
12     return(d.sort_index()[:n]) #返回排序后的前n行数据
13 def f_df2(d,k1):
14     return(d[k1])#返回分组后表的k1列，结果为series，层次化索引
15 print(df.groupby('key1').apply(f_df1,2),'\n')#引入自己创建的函数，注意书写格式，f_df1的第一个参数默认df 第二个参数,2的形式
16 print(df.groupby('key1').apply(f_df2,'data2'))
17 #直接运行f_df函数
18 #参数直接写在后面，也可以写为：.apply(f_df,n=2)

结果：
      data1     data2 key1 key2
0  0.253473  0.731034    a  one
1  0.531445  0.061414    a  two
2  0.733180  0.122293    b  one
3  0.550177  0.115940    b  two
4  0.101753  0.510581    a  one
           data1     data2 key1 key2
key1                                
a    0  0.253473  0.731034    a  one
     1  0.531445  0.061414    a  two
b    2  0.733180  0.122293    b  one
     3  0.550177  0.115940    b  two 
key1   
a     0    0.731034
      1    0.061414
      4    0.510581
b     2    0.122293
      3    0.115940
Name: data2, dtype: float64

 1 # 数据分组转换 .transform
 2 import numpy as np
 3 import pandas as pd
 4 df = pd.DataFrame({'data1':np.random.rand(5),
 5                    'data2':np.random.rand(5),
 6                    'key1':list('aabba'),
 7                    'key2':['one','two','one','two','one']})
 8 print(df)
 9 k_mean = df.groupby('key1').mean()
10 print(k_mean)
11 print(pd.merge(df,k_mean,left_on = 'key1',right_index = True).add_prefix('mean_'))#.add_profix('mean_'):增添前缀
12 print('------')
13 #通过分组，合并，得到一个包含均值的Dataframe   以key1位分组的列均值
14 
15 print(df.groupby('key2').mean())#按照key2分组求均值
16 print(df.groupby('key2').transform(np.mean))
17 #data1,data2每个位置的元素取对应分组列的均值
18 #字符串不能进行计算

结果：

data1 data2 key1 key2

0 0.845365 0.411704 a one

1 0.300226 0.411719 a two

2 0.476632 0.628493 b one

3 0.985675 0.304024 b two

4 0.418804 0.229940 a one

data1 data2

key1

a 0.521465 0.351121

b 0.731153 0.466258

mean_data1_x mean_data2_x mean_key1 mean_key2 mean_data1_y mean_data2_y

0 0.845365 0.411704 a one 0.521465 0.351121

1 0.300226 0.411719 a two 0.521465 0.351121

4 0.418804 0.229940 a one 0.521465 0.351121

2 0.476632 0.628493 b one 0.731153 0.466258

3 0.985675 0.304024 b two 0.731153 0.466258

------

data1 data2

key2

one 0.580267 0.423379

two 0.642951 0.357872

data1 data2

0 0.580267 0.423379

1 0.642951 0.357872

2 0.580267 0.423379

3 0.642951 0.357872

4 0.580267 0.423379