Split every row in df and add value to each element

Question

I have a df that looks like this:

user_index  movie_index  genre_index          cast_index
3590        1514         10|12|17|35          46|534
63          563          4|2|1|8              9|27

and was generated from:

import pandas as pd
ds = pd.DataFrame({'user_index': [3590,63], 'movie_index': [1514,563], 
'genre_index':['10|12|17|35', '4|2|1|8'], 'cast_index':['46|534', '9|27']})

I need to split every row by '|' (whereas converting every row to list) and to add to each element some value to get such df (here, '5' is added element-wise in column 'genre_index', '2' is added element-wise in column 'user_index'):

    user_index  movie_index  genre_index          cast_index
    [3592]      [1514]       [15,17,22,38]        [46,534]
    [65]        [563]        [9,7,6,13]            [9,27]

to achieve this, I create a function that takes column as an argument, splits it and adds a value element-wise (I don't take 'df' as argument as an added value would be different for each column) looks like this:

def df_convertion(input_series, offset):
    column = input_series.str.split('|', expand=False).apply(lambda x: x + offset)
    return (column)

but apparently the whole thing doesn't work as desired (I've tried for 'genre_index' column) and returns such an error:

TypeError: can only concatenate list (not "int") to list

Any help in fixing it would be very appreciated!

Answer 1

This is one of those rare times I'll suggest using apply. Try to see whether you can use some other form of representation for your data.

offset_dct = {'user_index': 2, 'genre_index': 5}
df = df.fillna('').astype(str).apply(lambda x: [
    [int(z) + offset_dct.get(x.name, 0) for z in y.split('|')] for y in x])

df
  cast_index       genre_index movie_index user_index
0  [46, 534]  [15, 17, 22, 40]      [1514]     [3592]
1    [9, 27]     [9, 7, 6, 13]       [563]       [65]