pandas: replace corresponding values in a comma separated column based on a list and another column-CodePudding

I have a dataframe and a list as follows:

import pandas as pd
import numpy as np

df = pd.DataFrame({'IDs':['d,f,o','d,f','d,f,o','d,f','d,f'],
                'Names':['APPLE ABCD ONE','date ABCD','NO foo YES','ORANGE AVAILABLE','TEA AVAILABLE']})

my_list = ['APPLE', 'ORANGE', 'LEMONS', 'STRAWBERRY', 'BLUEBERRY']

I would like to replace the comma separated values in the IDs column with the corresponding values from the Names column in case they appear in my_list.

desired output:
df.IDs => ['APPLE,f,o', 'd,f', 'd,f,o', 'ORANGE,f', 'd,f']

to find out whether the row contains the values in the list I have tried:

df['Names'].apply(lambda x: any([k in x for k in my_list]))

and to replace the values in the IDs column I have tried the following but I am not sure how to indicate that only the corresponding value should change,

df.IDs.apply(lambda i: i if i in my_list else 'don't know what to do here')

and i think I can use np.where() to perform the whole replacement based on conditions

np.where(df['Names'].apply(lambda x: any([k in x for k in my_list])) == True, df.IDs.apply(lambda i: i if i in my_list else 'don't know what to do here'), df.IDs)

CodePudding user response：

You could split/explode, then replace your values from the list, and agg back to the original shape:

(df.assign(IDs=df['IDs'].str.split(','),     # strings to lists
           Names=df['Names'].str.split(' ')
          )
   .apply(pd.Series.explode)                 # lists to rows
    # map the Names in place of Ids is in my_list
   .assign(IDs=lambda d: d['IDs'].mask(d['Names'].isin(my_list), d['Names']))
    # reshape back to original by joining
   .groupby(level=0).agg({'IDs': ','.join, 'Names': ' '.join})
)

output:

         IDs             Names
0  APPLE,f,o    APPLE ABCD ONE
1        d,f         date ABCD
2      d,f,o        NO foo YES
3   ORANGE,f  ORANGE AVAILABLE
4        d,f     TEA AVAILABLE