如何将group by的结果传递给Pivot？

Question

我的目标是将pivot函数应用于包含重复记录的数据帧。我通过在数据框中添加一个唯一的列来解决这个问题：

my_df['id_column'] = range(1, len(my_df.index)+1)

df_pivot = my_df.pivot(index ='id_column', columns = 'type', values = 'age_16_18').fillna(0).astype(int)

我想知道如何将pivot应用于数据帧，而不删除重复项或使用pivot\u表？首先按多列分组，然后将结果传递给pivot函数。我不知道分组后如何将结果传递给pivot。你知道吗

    year  category  state_name  type    is_state gender age_16_18 age_18_30
0   2001  Foreigners  CA       Convicts   0       M       8          5
1   2001  Indians     NY       Convicts   0       F       5          2 
2   2005  Foreigners  NY       Others     1       M       0          9
3   2009  Indians     NJ       Detenus    0       F       7          0

Answer 1

现在还不完全清楚你在尝试什么，但是看看你是否能从下面的方法中得到一些灵感。您希望按哪些列分组？你知道吗

import pandas
my_df = pandas.DataFrame( { 'year' : [2001, 2001, 2005, 2009] ,
                            'category' : ['Foreigners','Indians','Foreigners','Indians'] ,
                            'state_name': ['CA','NY','NY','NJ' ],
                            'type': ['Convicts', 'Convicts','Others','Detenus'],
                            'is_state' : [0,0,1,0] ,
                            'gender' : ['M','F','M','F'],
                            'age_16_18':[8,5,0,7],
                            'age_18_30' : [5,2,9,0] }, columns=[ 'year','category','state_name','type','is_state','gender','age_16_18','age_18_30'])

>>> my_df.pivot( columns = 'type', values = 'age_16_18' )
type  Convicts  Detenus  Others
0          8.0      NaN     NaN
1          5.0      NaN     NaN
2          NaN      NaN     0.0
3          NaN      7.0     NaN

>>> my_df['key'] = my_df.category.str.cat(my_df.gender)

>>> my_df.pivot( index='key', columns = 'type', values = 'age_16_18' )
type         Convicts  Detenus  Others
key
ForeignersM       8.0      NaN     0.0
IndiansF          5.0      7.0     NaN