将数据框的多行折叠为一行-基于唯一键 [英] Collapse mutiple rows of a dataframe into one row - based on a unique key

问题描述

我的数据框为:

1 A1
1 A11
2 A2
2 A22
2 A23
3 A3
3 A33
4 A4
4 A44
4 A444
5 A5

我需要的是:-

1 |  A1, A11 
2 | A2, A22, A23
3 | A3, A33
4 | A4, A44, A444 
5 | A5 

即.每列可以有不同数量的行.

ie. each column can have different number of rows present.

无论如何，我可以优雅地折叠它们，而无需使用字典中的读数，然后将其合并到适用的列表中.在传统意义上，我需要对此执行多个连接-可以吗?

Anyway I can collapse them elegantly, without using the reading from dict and then concat to the list as applicable. In the traditional sense I need to perform multiple joins on this - Any way around ?

请注意，最后只能有2列.

Note that there should be only 2 final columns.

推荐答案

df =pd.DataFrame({'A':[1,1,1,2,2,3,3,3],    'B':['aaa','bbb','cc','gg','aaa','bbb','cc','gg']})

def f(x):
    return [x['B'].values]

df.groupby('A').apply(f)

在要减少的列上创建一个分组依据，然后应用一个函数，该函数按每个分组的列表返回该分组的结果.请注意，这将返回一个序列.

Create a group by on the column you want to reduce over and then apply a function that returns the results of the group by an a list per group. Note this returns a series.

更新:将系列更改为数据框.

Update: change the series to a dataframe.

series =df.groupby('A').apply(f)
series.name = 'metric'
series.reset_index()

这篇关于将数据框的多行折叠为一行-基于唯一键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！