Replace pandas DataFrame column values with dict values looked up by separate column

Given a pandas DataFrame containing nan:

import numpy as np
import pandas as pd

bad_df = pd.DataFrame({'foo': [np.nan, 1.0, 2.0], 'bar': ['a', 'b', 'c']})

And also given a dictionary with lookup keys that may exist in the bar column:

replace_values = {'a': 7.0, 'd': 10.0}

How can the nan values in the foo column be replace with values in the replace_values dictionary based on the lookup of bar? The resulting DataFrame would look like:

expected_df = pd.DataFrame({'foo': [7.0, 1.0, 2.0], 'bar': ['a', 'b', 'c']})

The typical fillna and replace methods on DataFrames don’t seem to have this functionality since they will replace all na with the same value as opposed to a lookup value.

Thank you in advance for your consideration and response.

>Solution :

You can use loc with a condition using isna, and assign values using map:

bad_df.loc[bad_df['foo'].isna(), 'foo'] = bad_df['bar'].map(replace_values)

Or fillna with map as an alternative:

bad_df['foo'] = bad_df['foo'].fillna(bad_df['bar'].map(replace_values))

Both will ouput:

   foo bar
0  7.0   a
1  1.0   b
2  2.0   c