CODEWITHSUNDEEP
pandas
ComplicatedPhenomenon
ComplicatedPhenomenon

Reputation: 4199

make a transformation of dataframe

I need advice to transform a DataFrame df from

               timestamp   group        v1    v2    
0    11/14/2019 00:00:00      g1      0.00    1      
1    11/14/2019 00:00:00      g2      0.00    2

to 

          timestamp   new_group         v 
11/14/2019 00:00:00       g1_v1      0.00    
11/14/2019 00:00:00       g1_v2         1
11/14/2019 00:00:00       g2_v1      0.00    
11/14/2019 00:00:00       g2_v2         2

The logic is to combine v1 with v2 then make it as a new column as v, meanwhile, create a new column new_group which join the original group value and original column name(v1 or v2) as its value.

Upvotes: 0

Views: 32

Answers (1)

sammywemmy
sammywemmy

Reputation: 28729

melt the dataframe, combine the new columns created into one using str cat and filter for required columns

res = (df
       .melt(['timestamp','group'])
       .assign(new_group = lambda x: x.group.str.cat(x.variable,sep="_"))
       .filter(['timestamp','new_group','value'])
      )

res


       timestamp    new_group   value
0   11/14/2019 00:00:00 g1_v1   0.0
1   11/14/2019 00:00:00 g2_v1   0.0
2   11/14/2019 00:00:00 g1_v2   1.0
3   11/14/2019 00:00:00 g2_v2   2.0

Upvotes: 1

Related Questions