Level是有层次的,这样阅读非常方便,但是对编程定位比较麻烦. # 数据准备
import pandas as pdimport numpy as npdf = pd.DataFrame(np.arange(0, 14).reshape(7,2),columns =['a','b'] )df.a = df.a %3df['who'] = 'Bob'df.loc[df.a%4==0,'who'] = 'Alice'
a | b | who | |
---|---|---|---|
0 | 0 | 1 | Alice |
1 | 2 | 3 | Bob |
2 | 1 | 5 | Bob |
3 | 0 | 7 | Alice |
4 | 2 | 9 | Bob |
5 | 1 | 11 | Bob |
6 | 0 | 13 | Alice |
gp1 = df.groupby('who').agg({'b':[sum,np.max, np.min], 'a':sum})gp1
b | a | |||
---|---|---|---|---|
sum | amax | amin | sum | |
who | ||||
Alice | 8.0 | 7.0 | 1.0 | 0 |
Bob | 28.0 | 11.0 | 3.0 | 6 |
索引是有层次的,虚要通过下面这种方式,个人感觉不是很方便.下面介绍2种方法来解决这个问题
#有层次的索引访问方法gp1.loc['Bob', ('b', 'sum')]28.0
gp2 = gp1.copy(deep=True)gp2.columns = gp1.columns.droplevel(0)gp2
sum | amax | amin | sum | |
---|---|---|---|---|
who | ||||
Alice | 8.0 | 7.0 | 1.0 | 0 |
Bob | 28.0 | 11.0 | 3.0 | 6 |
gp3 = gp1.copy(deep=True)gp3.columns = ["_".join(x) for x in gp3.columns.ravel()]gp3
b_sum | b_amax | b_amin | a_sum | |
---|---|---|---|---|
who | ||||
Alice | 8.0 | 7.0 | 1.0 | 0 |
Bob | 28.0 | 11.0 | 3.0 | 6 |
以上这篇在Pandas中给多层索引降级的方法就是小编分享给大家的全部内容了,希望能给大家一个参考,也希望大家多多支持编程字典。
联系客服