Subtracting 2 columns of date

Question

I'm using python with pandas and datetime.

I have dataframe in csv file like that:

I too add new column called "Duration" by subtracting "Start date" from "End date".

what I have tried to do:

rides['Duration'] = end_time - start_time
duration = rides['Duration']

print(rides['Duration'])

and what I get is:

0     0 days 00:03:01
1     0 days 02:07:02
2     0 days 00:05:43

what I want to get is:

0     00:03:01
1     02:07:02
2     00:05:43

That is, I want to remove the "days" and show only the time.

Does this answer your question? [Remove the days in the timedelta object](https://stackoverflow.com/questions/53129971/remove-the-days-in-the-timedelta-object). — Henry Ecker, Jun 05 '21 at 17:40

score 0 · Answer 1 · answered Jun 05 '21 at 17:24

0

You can use % format specifier to extract hour, minute and second.

print("%d:%d:%d"%(rides['Duration'].hour,rides['Duration'].minute,rides['Duration'].second))

answered Jun 05 '21 at 17:24

Suchitra

Anurag Dabas · Answer 2 · 2021-06-05T18:13:52.287

0

Try via rsplit() if you want 'Duration' column of type object:

rides['Duration']=rides['Duration'].astype(str).str.rsplit(' ',1).str[1].str.strip('+|-')

OR

Try via rsplit() and to_datetime() if you want 'Duration' column of type datetime.time:

rides['Duration']=pd.to_datetime(rides['Duration'].astype(str).str.rsplit(' ',1).str[1].str.strip('+|-')).dt.time

edited Jun 05 '21 at 18:13

answered Jun 05 '21 at 17:28

Anurag Dabas

Thank you, your first method worked fine. but now I have another problem. The dtype is object and I need it to be "timedelta64[ns]". How can I handle it? – Amit Ben-David Jun 05 '21 at 17:48
then use the second method as it will create `datetime.time` – Anurag Dabas Jun 05 '21 at 18:04
I've tried the second method, I get this error: "ParserError: Unknown string format: +23:04:14" – Amit Ben-David Jun 05 '21 at 18:08
Thank you so much for your effort and help and sorry if I ask too many questions and bother you but the dtype is still "object" :( – Amit Ben-David Jun 05 '21 at 18:24
No.....If you used second method then it's of type `datetime.time` and you can cross verify by `rides['Duration'].map(type)` – Anurag Dabas Jun 05 '21 at 18:25

2 Answers2