Remove the days in Timedelta column

Question

This is a column in my data frame and I want to remove the 0 days so the column will just show the time

           Time
0     0 days 04:00:00
1     0 days 04:01:00
2     0 days 04:03:00
3     0 days 04:04:00
4     0 days 04:07:00

Desired output:

         Time
0      04:00:00
1      04:01:00
2      04:03:00
3      04:04:00
4      04:07:00

I've been recommended this question > Link

And tried to suit it to my code with this code:

df['Time'] = df['Time'] - pd.to_timedelta(df['Time'].dt.days, unit='d')

But I'm still getting the same output:

           Time
0     0 days 04:00:00
1     0 days 04:01:00
2     0 days 04:03:00
3     0 days 04:04:00
4     0 days 04:07:00

If you want a custom string representation of timedelta, you'll have to [format timedelta to string](https://stackoverflow.com/q/538666/10197418) by yourself. There's no such thing as `strftime` as you have it for formatting datetime to string. — FObersteiner, Dec 30 '21 at 15:46

Paul · Accepted Answer · 2021-12-30T15:49:25.650

1

If you don't care about the time-intelligence of dtype: timedelta64[ns] you can cast it to string and remove the '0 days ' part:

df.Time.astype(str).str.replace('0 days ', '')

In case the Time column is not always '0 days', you can remove the days with the following code:

df['Time'] = df['Time'] - pd.to_timedelta(df['Time'].dt.days, unit='d')

edited Dec 30 '21 at 15:49

answered Dec 30 '21 at 15:39

Paul

What if the days attribute is not zero? – FObersteiner Dec 30 '21 at 15:46
Since it is a time column, I assume it is always 0. However, if this is not the case:`df['Time'] = df['Time'] - pd.to_timedelta(df['Time'].dt.days, unit='d')` – Paul Dec 30 '21 at 15:47
With the data I'm collecting it will always be zero. @ MrFuppes – NoLimitLondon Dec 30 '21 at 15:50
Then to represent time, OP should use datetime data type, not timedelta – FObersteiner Dec 30 '21 at 15:52

score 0 · Answer 2 · answered Dec 30 '21 at 16:14

0

Try with

df['Time'].astype(str).str.split(' ').str[-1]
Out[288]: 
0    04:00:00
1    04:01:00
2    04:03:00
3    04:07:00
Name: Time, dtype: object

answered Dec 30 '21 at 16:14

BENY

2 Answers2