How to change NaN values to other values depending on columny type

Question

I am trying to change NaN values in specific types of columns of a pandas dataframe.

I would like to do this with an if statement where I can use the isna() function and substitute NaN with text for the object type column and NaN with 0 values for the float64 type column.

Answer 1

You can use select_dtypes to get the wanted type, and fillna+update to updated the NAs/NaNs:

df = pd.DataFrame({'col1': [1,np.nan,3],
                   'col2': ['a',np.nan,'c'],
                   'col3': pd.Series(['a', pd.NA, 'c'], dtype='string')})
df.update(df.select_dtypes('number').fillna(-1))
df.update(df.select_dtypes('object').fillna('empty'))
df.update(df.select_dtypes('string').fillna('empty2'))

NB. check the types first with df.dtypes, in recent pandas versions it is possible to have 'string' as type for strings instead of 'object'

output:

   col1   col2    col3
0   1.0      a       a
1  -1.0  empty  empty2
2   3.0      c       c

input:

   col1 col2  col3
0   1.0    a     a
1   NaN  NaN  <NA>
2   3.0    c     c