Using rolling with datetime index and duplicate timestamps

Asked Aug 09 '21 at 16:46

Active Aug 09 '21 at 16:46

Viewed 90 times

I have a pandas dataframe df that goes like this

index     data
midnight  1
1am       1
2am       1
2am       1

I want to count the number of lines in the last 90 minutes excl, to do so, I do df.rolling("90min", closed="left").fillna(0)

Result is

index     data
midnight  0
1am       1
2am       1
2am       2

Considering 2am is a duplicate index, I would like the last value to be 1 instead of 2. Any idea ?

asked Aug 09 '21 at 16:46

LotsOfQuestions

so how you are performing rolling on the string data? – Anurag Dabas Aug 09 '21 at 17:32
try [removing duplicates](https://stackoverflow.com/questions/13035764/remove-pandas-rows-with-duplicate-indices) first and then roll – woblob Aug 10 '21 at 07:13
@woblob I don't want to remove duplicate. each line is associated with an id and I want the rolling count at every given moment to be linked with the id – LotsOfQuestions Aug 11 '21 at 08:33
@AnuragDabas I'm performing rolling on the data field, which is numeric – LotsOfQuestions Aug 11 '21 at 08:34

0 Answers0