Stack pandas numeric column with a new presence/absence column

Question

I am trying to stack two columns in one column, but I want to keep a boolean field that denotes where the new values coming from. Using pivot and melt did not yield much success. Any help would be useful. Here is a small example

param | accuracy | accuracy_with_reg
----------------------------
4     |  0.813    | 0.934
5     |  0.456    | 0.654
6     |  0.342    | 0.564

to the following

param | accuracy | regularized 
----------------------------
4     |  0.813    | False
5     |  0.456    | False
6     |  0.342    | False
4     |  0.934    | True
5     |  0.654    | True
6     |  0.654    | True

Thanks

Ben.T · Accepted Answer · 2021-02-18 19:44:58Z

You can use melt like

pd.melt(df, id_vars=['param'], value_vars=['accuracy', 'accuracy_with_reg'],
        var_name='regularized', value_name='accuracy_val')

   param        regularized  accuracy_val
0      4           accuracy         0.813
1      5           accuracy         0.456
2      6           accuracy         0.342
3      4  accuracy_with_reg         0.934
4      5  accuracy_with_reg         0.654
5      6  accuracy_with_reg         0.564

if you want True or False in the column regularized, then you have different method but I would go on rename the columns before

pd.melt(df.rename(columns={'accuracy':False, 'accuracy_with_reg':True}), 
        id_vars=['param'], value_vars=[False, True],
        var_name='regularized', value_name='accuracy')
   param regularized  accuracy
0      4       False     0.813
1      5       False     0.456
2      6       False     0.342
3      4        True     0.934
4      5        True     0.654
5      6        True     0.564

Collectives™ on Stack Overflow

Stack pandas numeric column with a new presence/absence column

1 Answer 1

Comments

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Related