Python: Pandas df.fillna() function change all data type into object

Question

I want to fill feature with null value in dataframe. But when I fill to all feature, every data type I'm filling was changed to "Object".

I have dataframe with data type:

 #   Column         Non-Null Count  Dtype  
---  ------         --------------  -----  
 0   umur           7832 non-null   float64
 1   jenis_kelamin  7840 non-null   object 
 2   pekerjaan      7760 non-null   object 
 3   provinsi       7831 non-null   object 
 4   gaji           7843 non-null   float64
 5   is_menikah     7917 non-null   object 
 6   is_keturunan   7917 non-null   object 
 7   berat          7861 non-null   float64
 8   tinggi         7843 non-null   float64
 9   sampo          7858 non-null   object 
 10  is_merokok     7917 non-null   object 
 11  pendidikan     7847 non-null   object 
 12  stress         7853 non-null   float64

And I use fillna() for filling null value to every feature

# Feature categoric type inputation
df['jenis_kelamin'].fillna(df['jenis_kelamin'].mode()[0], inplace = True)
df['pekerjaan'].fillna(df['pekerjaan'].mode()[0], inplace = True)
df['provinsi'].fillna(df['provinsi'].mode()[0], inplace = True)
df['sampo'].fillna(df['sampo'].mode()[0], inplace = True)
df['pendidikan'].fillna(df['pendidikan'].mode()[0], inplace = True)

# Feature numeric type inputation
df['umur'].fillna(df['umur'].mean, inplace = True) 
df['gaji'].fillna(df['gaji'].mean, inplace = True) 
df['berat'].fillna(df['berat'].mean, inplace = True)
df['tinggi'].fillna(df['tinggi'].mean, inplace = True)
df['stress'].fillna(df['stress'].mean, inplace = True)

But after that, all feature's data type has been changed to Object:

 #   Column         Non-Null Count  Dtype  
---  ------         --------------  -----  
 0   umur           7917 non-null   object 
 1   jenis_kelamin  7917 non-null   object 
 2   pekerjaan      7917 non-null   object 
 3   provinsi       7917 non-null   object 
 4   gaji           7917 non-null   object 
 5   is_menikah     7917 non-null   object 
 6   is_keturunan   7917 non-null   object 
 7   berat          7917 non-null   object 
 8   tinggi         7917 non-null   object 
 9   sampo          7917 non-null   object 
 10  is_merokok     7917 non-null   object 
 11  pendidikan     7917 non-null   object 
 12  stress         7917 non-null   object 

I think it can be work to convert every feature with astype(), but is there any other efficient way to fill null value without change the datatype?

Answer 1

I think you are missing your brackets on .mean(), so it is filling the series with a method instead of the actual values.

You want, for example:

df['umur'].fillna(df['umur'].mean(), inplace = True)