I have a problem using the round() funktion on a pandas.Series / DataFrame filled with Timestamps.
Code example to round to the nearest 10-second mark:
import pandas as pd
date = pd.DataFrame([pd.to_datetime('2022-03-02 06:46:05'), pd.to_datetime('2022-03-02 06:46:15'), pd.to_datetime('2022-03-02 06:46:25'), pd.to_datetime('2022-03-02 06:46:35'), pd.to_datetime('2022-03-02 06:46:45'), pd.to_datetime('2022-03-02 06:46:55'), pd.to_datetime('2022-03-02 06:47:05'), pd.to_datetime('2022-03-02 06:47:15'), pd.to_datetime('2022-03-02 06:47:25')])
date[1] = date[0].round('10s')
date
OUT:
0 1
0 2022-03-02 06:46:05 2022-03-02 06:46:00
1 2022-03-02 06:46:15 2022-03-02 06:46:20
2 2022-03-02 06:46:25 2022-03-02 06:46:20
3 2022-03-02 06:46:35 2022-03-02 06:46:40
4 2022-03-02 06:46:45 2022-03-02 06:46:40
5 2022-03-02 06:46:55 2022-03-02 06:47:00
6 2022-03-02 06:47:05 2022-03-02 06:47:00
7 2022-03-02 06:47:15 2022-03-02 06:47:20
8 2022-03-02 06:47:25 2022-03-02 06:47:20
dtype: datetime64[ns]
Whenever a Timestamp has a seconds value in [5, 25, 45] the rounded value is set to [0, 20, 40], although it should be set to [10, 30, 50]. Any idea on how to fix this?
Thanks in advance!
>Solution :
Use trick – add some small timedelta, because python should round 5 not like expected:
date[1] = date[0].add(pd.Timedelta('1us')).round('10s')
print (date)
0 1
0 2022-03-02 06:46:05 2022-03-02 06:46:10
1 2022-03-02 06:46:15 2022-03-02 06:46:20
2 2022-03-02 06:46:25 2022-03-02 06:46:30
3 2022-03-02 06:46:35 2022-03-02 06:46:40
4 2022-03-02 06:46:45 2022-03-02 06:46:50
5 2022-03-02 06:46:55 2022-03-02 06:47:00
6 2022-03-02 06:47:05 2022-03-02 06:47:10
7 2022-03-02 06:47:15 2022-03-02 06:47:20
8 2022-03-02 06:47:25 2022-03-02 06:47:30