在groupby之后选择列而不使用显式列名 [英] Selecting column after groupby without using explicit column name

问题描述

具有以下数据集:

import pandas as pd
df = pd.DataFrame({'Date':['26-12-2018','26-12-2018','27-12-2018','27-12-2018','28-12-2018','28-12-2018'],
                   'In':['A','B','D','Z','Q','E'],
                   'Out' : ['Z', 'D', 'F', 'H', 'Z', 'A'],
                   'Score_in' : ['6', '2', '1', '0', '1', '3'], 
                   'Score_out' : ['2','3','0', '1','1','3'],
                   'Place' : ['One','Two','Four', 'Two','Two','One']})

我想对通用规则的groupby规则进行编码，以尝试对子集的创建进行参数化. 例如，代替以下内容:

I would like to code groupby rules on a generic form in order to try parameterizing subsets creation. For instance, instead of the following:

df.groupby('In').Score_in.sum()

我想我想要的输出将是诸如#1或#2之类的，具有df.columns[]或.iloc[:,[]]语法，例如:

I suppose my desired output would be something like #1 or #2 with df.columns[] or .iloc[:,[]] syntaxes like:

df.groupby(df.columns[1]).df.iloc[:,[3]].sum() #1
df.groupby(df.iloc[:,[0]]).df.iloc[:,[3]].sum() #2

当然，以上语法都不起作用.有帮助吗?

Of course, none of the above syntaxes works. Any help?

推荐答案

实际上，问题不是出在groupby上，而是关于以后如何保留特定列的问题. groupby没有df属性，因此无法通过这种方式工作.

Actually the problem is not with the groupby, it's about how you keep a particular column afterwards. groupby has no df attribute, so it can't work this way.

这是一段可以按预期工作的代码:

Here is a piece of code that works as expected:

df.groupby(df.columns[1])[df.columns[3]].sum()

In  Score_in
A   6
B   2
D   1
E   3
Q   1
Z   0

注意:我将Score_in和Score_out强制转换为整数，否则groupby无法正常工作.

Notice: I casted Score_in and Score_out as integers or else the groupby would'nt work.

这篇关于在groupby之后选择列而不使用显式列名的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！