Shifting every second value in a column to a new column for each unique index in pandas-CodePudding

I have a large pandas dataframe as follows.

Input df

member_id   attribute1
1           AM
1           A
26          TSE
26          TSA
26          TSE
      ----
97736       TMA      
97736       TTE
97736       TMA
97736       ALM

Is there any way I can efficiently shift every second value in attribute1 column to a new column lets say attribute2 for each unique member_id index.

Output df

member_id   attribute1   attribute2
1           AM            A
26          TSE           TSA
26          TSA           TSE
   
      ----
97736       TMA           TTE     
97736       TTE           TMA
97736       TMA           ALM

CodePudding user response：

Try:

# shift the rows up
df['attribute2'] = df.groupby('member_id')['attribute1'].shift(-1)

# drop the last row for each `member_id`
df = df[df['member_id'].duplicated(keep='last')]

Output:

   member_id attribute1 attribute2
0          1         AM          A
2         26        TSE        TSA
3         26        TSA        TSE
5      97736        TMA        TTE
6      97736        TTE        TMA
7      97736        TMA        ALM