从数据框列中选择非None值

Question

I would like to use the fillna function to fill None value of a column with its own first most frequent value that is not None or nan.

Input DF:

Col_A
a
None
None
c
c
d
d

The output Dataframe could be either:

Col_A
a
c
c
c
c
d
d

Any suggestion would be very appreciated. Many Thanks, Best Regards, Carlo

Answer 1

Prelude: If your None is actually a string , you can simplify any headaches by getting rid of them first-up. Use replace :

df = df.replace('None', np.nan)

---

I believe you could use fillna + value_counts :

df

  Col_A
0     a
1   NaN
2   NaN
3     c
4     c
5     d
6     d

df.fillna(df.Col_A.value_counts(sort=False).index[0])

  Col_A
0     a
1     c
2     c
3     c
4     c
5     d
6     d

Or, with Vaishali's suggestion, use idxmax to pick c :

df.fillna(df.Col_A.value_counts(sort=False).idxmax())

  Col_A
0     a
1     c
2     c
3     c
4     c
5     d
6     d

The fill-values could either be c or d , depending on whether you include sort=False or not.

Details

df.Col_A.value_counts(sort=False)

c    2
a    1
d    2
Name: Col_A, dtype: int64

Answer 2

fillna + mode

df.Col_A.fillna(df.Col_A.mode()[0])
Out[963]: 
0    a
1    c
2    c
3    c
4    c
5    d
6    d
Name: Col_A, dtype: object

Answer 3

To address 'None', you need to use replace then fillna much like @COLDSPEED suggests:

dr = df.Col_A.replace('None',np.nan)
dr.fillna(dr.dropna().value_counts().index[0])

Output:

0    a
1    d
2    d
3    c
4    c
5    d
6    d
Name: Col_A, dtype: object