[FIXED] Use list to map subset of dataframe columns to dictionary values

Issue

Is there a way to pass a list of columns to map a dictionary's values in a single call?

I can do it by stating each column but wondering if I can shorten up the syntax.

Looked at this post for reference.

Sample data:

import pandas as pd
 
# initialize data of lists.
data = {'id': ['a', 'b', 'c', 'd', 'e'],
        'col1': ['PHX', 'BKN', 'X', 'PHX', 'X'],
        'col2': ['X', 'PHX', 'BKN', 'BKN', 'X'],
        'col3': ['PHX', 'BKN', 'PHX', 'BKN', 'PHX']
       }
 
df = pd.DataFrame(data)
df

  id col1 col2 col3
0  a  PHX    X  PHX
1  b  BKN  PHX  BKN
2  c    X  BKN  PHX
3  d  PHX  BKN  BKN
4  e    X    X  PHX

I want to apply this mapping to col1 and col2 but not col3:

name_dict = {'PHX' : 'PHO', 'BKN': 'NJN'}

So PHX -> PHO and BKN -> NJN.

This replaces all cols but I want a subset of columns:

df = df.replace(name_dict)

Daisy-chaining 'replace' works:

df = df.replace({'col1': name_dict}).replace({'col2': name_dict})

This also works:

df = df.replace({'col1': name_dict, 'col2': name_dict})

Can I shorten this up a bit, something like:

df = df.replace({['col1', 'col2']: name_dict})

However, this results in TypeError: unhashable type: 'list'.

Desired output (col1 and col2 values updated but not col3):

  id col1 col2 col3
0  a  PHO    X  PHX
1  b  NJN  PHO  BKN
2  c    X  NJN  PHX
3  d  PHO  NJN  BKN
4  e    X    X  PHX

Solution

Using dict.fromkeys:

df.replace(dict.fromkeys(['col1', 'col2'], name_dict))

Or, if you want to modify in place, with a list of columns:

cols = ['col1', 'col2']
df[cols] = df[cols].replace(name_dict)

Output:

  id col1 col2 col3
0  a  PHO    X  PHX
1  b  NJN  PHO  BKN
2  c    X  NJN  PHX
3  d  PHO  NJN  BKN
4  e    X    X  PHX

Answered By - mozway