Rolling count specific number

Question

Suppose I have dataframe "df" like this

Time	Group	Data
2022-10-01 00:05:00	A	0
2022-10-01 00:10:00	A	0
2022-10-01 00:15:00	A	1
2022-10-01 00:20:00	A	1
2022-10-01 00:25:00	A	1
2022-10-01 00:30:00	A	0
2022-10-01 00:35:00	A	1
2022-10-01 00:40:00	A	0
2022-10-01 00:05:00	B	11
2022-10-01 00:10:00	B	0
2022-10-01 00:15:00	B	12
2022-10-01 00:20:00	B	13
2022-10-01 00:25:00	B	0
2022-10-01 00:30:00	B	0
2022-10-01 00:35:00	B	15
2022-10-01 00:40:00	B	16

Assume That I already sort out data by Group and Time already I would love to count occurence of 0 in previous 15 minutes include itself

Which result should be like

Time	Group	Data	Count_0_last_15_min
2022-10-01 00:05:00	A	0	1
2022-10-01 00:10:00	A	0	2
2022-10-01 00:15:00	A	1	2
2022-10-01 00:20:00	A	1	1
2022-10-01 00:25:00	A	1	0
2022-10-01 00:30:00	A	0	1
2022-10-01 00:35:00	A	1	1
2022-10-01 00:40:00	A	0	2
2022-10-01 00:05:00	B	11	0
2022-10-01 00:10:00	B	0	1
2022-10-01 00:15:00	B	12	1
2022-10-01 00:20:00	B	13	1
2022-10-01 00:25:00	B	0	1
2022-10-01 00:30:00	B	0	2
2022-10-01 00:35:00	B	0	3
2022-10-01 00:40:00	B	16	2

currently I try to use rolling to get data from each from previous record

df.groupby('Group')['Data'].rolling(3,min_periods=1)

however I stuck after rolling part to only counting "0" occurrence ( I did try .eq(0).sum() but it can't apply with rolling() method

is there another method to group by and rolling count for this solution?

Answer 1

You can use:

(df.assign(Count_0_last_15_min=df['Data'].eq(0))
   .groupby('Group', as_index=False)
   .rolling(3, min_periods=1)['Count_0_last_15_min'].sum()
   .droplevel(0)
)

Output:

0     1.0
1     2.0
2     2.0
3     1.0
4     0.0
5     1.0
6     1.0
7     2.0
8     0.0
9     1.0
10    1.0
11    1.0
12    1.0
13    2.0
14    2.0
15    1.0
Name: Count_0_last_15_min, dtype: float64