Fill missing dates with another column value

Question

I've been trying to do something that I thought would be simple, but I'm facing an issue I'm not understanding. I have two columns: date_published and date_obtained , of course, I've got all data for date_obtained , but not for date_published . My approach was to fill the missing date_published with date obtained minus 1 day (might also be the median difference, but I'll ignore that).

df looks like this:

date_published    date_obtained
 2017-12-20        2017-12-22
    NaT            2017-12-23

And should look like this afterwards:

date_published    date_obtained
 2017-12-20        2017-12-22
 2017-12-22        2017-12-23

I tried the following:

date_delta=(Df.date_obtained-datetime.timedelta(days=1))
Df.loc['date_published']=Df.date_published.fillna((date_delta))

But, to my surprise, that didn't fill any NaT , and also created a missing value in all of my data frame. Also tried filling with just Df.date_obtained , but the result was the same. What am I missing?

Answer 1

You were almost there, you should've either done:

u = df.date_obtained - pd.Timedelta(days=1)
df['date_published'] = df.date_published.fillna(u)

Or,

df.loc[:, 'date_published'] = df.date_published.fillna(u)

Using loc to refer to the columns (otherwise, it will try to check the index for the label you passed).