How to pivot and extend dataframe columns

Question

I have the following dataframe:

 name      precision      recall
    a         0.28        0.23    
    b         1.00        0.00    
    a         0.31        0.23    
    b         0.25        0.00

The desired output is:

a_precision   a_recall    b_precision   b_recall
   0.28         0.23        1.00         0.00
   0.31         0.23        0.25         0.00

Any idea how to perform this pivot-like operation?

In my dataset I have 5 different names (a,b,c,d,e) and each sixth row it starts again with name a, b… and so on. Beside of precision and recall I have another column called f1_score. So probably the solution should be adaptable to a different dataframe schema.

I am looking forward how you would tackle this problem.

Asked By: Daniel

||

Source

Answer 1

You can use pivot with a little bit of reworking your dataframe:

df2 = (df.assign(group=df.groupby('name').cumcount()) # get position across groups
         .pivot(index='group', columns='name')
      )
df2.columns = ['%s_%s' % (b,a) for (a,b) in df2.columns]
df2.sort_index(axis=1)

output:

       a_precision  a_recall  b_precision  b_recall
group                                              
0             0.28      0.23         1.00       0.0
1             0.31      0.23         0.25       0.0

Answered By: mozway

Answer 2

def function1(dd:pd.DataFrame):
    dd1=dd.iloc[:,1:].add_prefix(dd.name.max()+"_").reset_index(drop=True)
    return dd1

pd.concat(pd.DataFrame(df1.groupby("name"))[1].map(function1).tolist(),axis=1)

out：

 a_precision  a_recall  b_precision  b_recall
0         0.28      0.23         1.00       0.0
1         0.31      0.23         0.25       0.0

Answered By: G.G

How to pivot and extend dataframe columns

Question:

Answers: