Exact matches in time-based .rolling() #13965
Description
I would like to allow_exact_matches in time-based .rolling(). Using an example similar to the one found in the documentation:
df = pd.DataFrame({'B': [1, 1, 2, np.nan, 4]},
index = [pd.Timestamp('20130101 09:00:00'),
pd.Timestamp('20130101 09:00:02'),
pd.Timestamp('20130101 09:00:03'),
pd.Timestamp('20130101 09:00:05'),
pd.Timestamp('20130101 09:00:06')])
I can compute the three-second rolling sum:
In [200]: df
Out[200]:
B
2013-01-01 09:00:00 1.0
2013-01-01 09:00:02 1.0
2013-01-01 09:00:03 2.0
2013-01-01 09:00:05 NaN
2013-01-01 09:00:06 4.0
In [201]: df.rolling('3s').sum()
Out[201]:
B
2013-01-01 09:00:00 1.0
2013-01-01 09:00:02 2.0
2013-01-01 09:00:03 3.0
2013-01-01 09:00:05 2.0
2013-01-01 09:00:06 4.0
But note that 09:00:03 does not include 09:00:00, 09:00:05 does not include 09:00:02, and 09:00:06 does not include 09:00:03. If I were to include these timestamps, then I would have:
B
2013-01-01 09:00:00 1.0
2013-01-01 09:00:02 2.0
2013-01-01 09:00:03 4.0
2013-01-01 09:00:05 3.0
2013-01-01 09:00:06 6.0
A quick-and-dirty way of getting these values is with:
from itertools import starmap
window = pd.Timedelta('3s')
begins = np.searchsorted(df.index, df.index - window)
ends = np.arange(len(df)) + 1
ranges = starmap(range, zip(begins, ends))
pd.concat([df.ix[r].sum() for r in ranges])
pd.merge_asof() has the parameter allow_exact_matches to permit this. Would it be possible to add this to .rolling()?
I can try this one myself.