.iloc & fillna - inplace=True vs inplace=False

Asked Apr 18 '19 at 10:23

Active Apr 18 '19 at 10:23

Viewed 705 times

Just out of curiosity I would like to have a quick explanation of the following "issue".

When I did this:

data_ogd.iloc[:, 3:10].fillna(value=0, inplace=True)

nothing changed at my dataframe columns and I still had the NAs at these columns.

But when I did this:

data_ogd.iloc[:, 3:10] = data_ogd.iloc[:, 3:10].fillna(value=0)

The NAs were filled with 0s.

Why was this happening?

asked Apr 18 '19 at 10:23

Outcast

You can check [this](https://stackoverflow.com/a/21463854/2901002) for explanation, same principe - chain assignment of Series, sometimes warning, sometimes not. – jezrael Apr 18 '19 at 10:25
Thanks @jezrael . Ah ok yes it is this common issue. By the way, I did not receive any warning this time. – Outcast Apr 18 '19 at 10:28
It depends of version of pandas, I guess in old version it should be warning. – jezrael Apr 18 '19 at 10:28
1

@jezrael, ok cool. Thank you! – Outcast Apr 18 '19 at 10:29
Sometimes it works as well (when I tried with version 0.23.0) – Mortz Apr 18 '19 at 10:30

0 Answers0