How to create a function which replace an empty values with the most appearing value or average value based on the specific columns

A csv file contains more than 18 columns and I want to perform data cleaning without dropping rows with empty values.

Some columns have string while other have float and int so I want to create function that will automatically check a certain columns with an empty value and replace that empty with either the most appearing value if this is string based column and the average value for int and float based columns instead of doing it manually.

|response|score|
|A       |25   |
|A       |     |
|B       |20   |
|C       |15   |
|        |25   |

My sample codes

df['response']=df['response'].fillna('A').astype(str)
df['score']=df['score'].fillna(df.score.mean()).astype(float)

You can try this:

df['score'] = df['score'].fillna(df.groupby('response')['score'].transform('mean'))

This will replace the nulls with mean of values of the same 'response'.

You can change the function to median, max, min - depends on your use case

If you want to do that for all the columns:

for col in df.columns:
    df[col] = df[col].fillna(df.groupby('response')[col].transform('mean'))

Use a custom function with pandas.api.types.is_numeric_dtype:

def filler(s):
    if pd.api.types.is_numeric_dtype(s):
        fill_value = s.mean()
    else:                         # add other conditions if needed
        fill_value = s.mode()[0]
    return s.fillna(fill_value)

out = df.apply(filler)

As a one-liner

out = df.apply(lambda s: s.fillna(s.mean() if pd.api.types.is_numeric_dtype(s)
                                  else s.mode()[0]))

Output:

  response  score
0        A  25.00
1        A  21.25
2        B  20.00
3        C  15.00
4        A  25.00