Pandas：仅在分组数据框中的选定列中连接值的有效方法

Question

I have a df such that

   LAST_MOD_DATE       ID    TITLE          TXT_ID             TXT  
0  1486047205463        2    TITLE-2        7                  ABC   
1  1486047205463        2    TITLE-2        5                  XYZ   
2  1486047205463        2    TITLE-2        6                  MNQ 

I would like to group it by ID so as to flatten it into a single row. The fields with differing values TXT_ID and TXT will be combined into one with comma separated values. So, Something like below:

ID 
2  1486047205463     TITLE-2        7, 5, 6          ABC, XYZ, MNQ

I am able to just get a single columns out by

df.groupby('ID')['TXT'].apply(lambda x:', '.join(x))

But how to do it on the entire df so that I can selectively join some columns while just choose the top values of the other columns within the same groups. Right now I am doing it by aggregating the values as a set and then expanding the set for some columns. But this doesn't seem very efficient

Answer 1

Use agg and supply what function you want to apply for each column. Here I give you a mixed example where I group only againt 'ID' to illustrate how to take the first element on 'TITLE', but you could group against it to for your sample (which might not be the general case you have:

df.groupby('ID').agg({'TITLE':'first', 
                      'TXT_ID':lambda x:', '.join(x),
                      'TXT':lambda x:', '.join(x)})
Out[288]: 
      TITLE   TXT_ID            TXT
ID                                 
2   TITLE-2  7, 5, 6  ABC, XYZ, MNQ