熊猫根据groupby计数获取列的值

Question

I have a dataframe named test like so

   ALT_K1 ALT_K2  ALT_K3  HS  VS
    1       A       1      45   2
    1       A       1      32  32
    1      1-1      70     1   1
    1      1-1      70     0   9
    1      A        2      3   0

and I groupby the first three columns and calculate the frequency of occurence like so

test_frequency = test.groupby(['ALT_K1', 'ALT_K2', 'ALT_K3']).size().reset_index(name='count')

I want to be able to get the values of the columns HS and VS given the number of times the combination of the three columns appears. For example, for the combination (1, A, 1) I want the to get the values of HS [45, 32] and VS [2, 32]

Being stuck in this one for two days now and would appreciate any help.

Thanks

Answer 1

I think you need custom lambda function with apply and unique :

test_frequency = test.groupby(['ALT_K1', 'ALT_K2', 'ALT_K3'])
                     .apply(lambda x: pd.Series([x['HS'].unique(),
                                                 x['VS'].unique()], index=['HS','VS']))
                     .reset_index()
print (test_frequency)

print (test_frequency)
   ALT_K1 ALT_K2  ALT_K3        HS       VS
0       1    1-1      70    [1, 0]   [1, 9]
1       1      A       1  [45, 32]  [2, 32]
2       1      A       2       [3]      [0]