Create new column in pandas dataframe that merges specific column names and corresponding values

Question

I have a dataframe that looks like the following:

       Colm1 ColmX Colm2
  0      1     2     3
  1      4     5     6

I need a new one as follows:

       Colm1 ColmX Colm2 Colm3
  0      1     2     3   Colm1_1_Colm2_3
  1      4     5     6   Colm1_4_Colm2_6

The merged value in Colm3 is constructed as an underscore separated list of pair of a specific list of columns, in this case, [Colm1, Colm2]

How do I go about doing this? I have a list of column names that I need to merge as above, to start off with. Thank you!

Answer 1

A stupid solution:

In [62]: df['Colm3'] = 'Colm1_'+df['Colm1'].astype(str)+'_Colm2_'+df['Colm2'].astype(str)

In [63]: df
Out[63]:
   Colm1  ColmX  Colm2            Colm3
0      1      2      3  Colm1_1_Colm2_3
1      4      5      6  Colm1_4_Colm2_6

A bit more generic solution:

cols=['Colm1','Colm2']

df['new'] = df[cols].apply(lambda x: x.name+'_'+x.astype(str)).add('_').sum(1).str.rstrip('_')

Detailed:

In [4]: df[cols].apply(lambda x: x.name+'_'+x.astype(str))
Out[4]:
     Colm1    Colm2
0  Colm1_1  Colm2_3
1  Colm1_4  Colm2_6

In [5]: df[cols].apply(lambda x: x.name+'_'+x.astype(str)).add('_')
Out[5]:
      Colm1     Colm2
0  Colm1_1_  Colm2_3_
1  Colm1_4_  Colm2_6_

In [6]: df[cols].apply(lambda x: x.name+'_'+x.astype(str)).add('_').sum(1)
Out[6]:
0    Colm1_1_Colm2_3_
1    Colm1_4_Colm2_6_
dtype: object

In [7]: df[cols].apply(lambda x: x.name+'_'+x.astype(str)).add('_').sum(1).str.rstrip('_')
Out[7]:
0    Colm1_1_Colm2_3
1    Colm1_4_Colm2_6
dtype: object