pd.fillna() not working on all numeric columns

Asked Mar 12 '19 at 14:42

Active Mar 12 '19 at 14:42

Viewed 36 times

I have a dataframe that has about 15 numeric columns, and for some reason when I pass those numeric columns as a list into the pd.fillna() method, 3 of them don't have their values filled, and I'm not sure why this is the case.

MY CODE:

num_cols = df.select_dtypes(include=[np.number]).columns.tolist()

num_cols then looks like this:

Then I pass num_cols into the dataframe using the fillna method, but some of the columns aren't affected by it.

df[num_cols].fillna(df[num_cols].median(), inplace=True)
df[num_cols].isnull().sum()

This last command produces the following picture:

Any idea why this wouldn't be working?

asked Mar 12 '19 at 14:42

Jonathan Bechtel

3,497
4
43
73

I think need `df[num_cols] = df[num_cols].fillna(df[num_cols].median())` – jezrael Mar 12 '19 at 14:45
@jezrael I just did this, and it worked on all the columns except one, which had dtype np.float16, which was not the same as the others. – Jonathan Bechtel Mar 12 '19 at 14:48
hmmm, interesting, what is your pandas version? – jezrael Mar 12 '19 at 14:50

pd.fillna() not working on all numeric columns

0 Answers0