pandas dataframe generate rows

April 22, 2023

I have a Input dataframe:

import pandas as pd

# Define the input data
data = {
    'ID':  [500, 200, 300],
    'A': [3, 3, ''],
    'B': [3, 1, ''],
    'C': [2, '' ,''],
    'D': ['', 2, 1],
    'E': ['', '',2 ],
}

# Convert the input data to a Pandas DataFrame
df = pd.DataFrame(data)

Input table

I need to transform this input as you can see in below output example:

If you have any ideas please share. Thank you very much!

>Solution :

You can repeat the rows using the max number per row, then compare to a groupby.cumcount to map the 1/'':

tmp = df.set_index('ID').apply(pd.to_numeric, errors='coerce')

out = (
 tmp.loc[tmp.index.repeat(tmp.max(axis=1))]
    .pipe(lambda d: d.gt(d.groupby(level='ID').cumcount(), axis=0))
    .replace({True: 1, False: ''})
    .reset_index()
)

NB. you can simplify the code if you use NaN in place of empty strings.

Output:

    ID  A  B  C  D  E
0  500  1  1  1      
1  500  1  1  1      
2  500  1  1         
3  200  1  1     1   
4  200  1        1   
5  200  1            
6  300           1  1
7  300              1