Expanding Window(time series forecasting) in R

Hi i am new to stackoverflow as well as R. I am currently taking a course on Machine Learning & Deep Learning using Rstudio & Python . In this course he is teaching Forecasting too, where he is using only Python to implement those code. In Feature Engineering part of Forecasting, he is implementing Expanding Window feature, which is part of Pandas. Can someone help me find this code in R.

The code he use in python is Feature['Expand_Max']=df['births'].expanding().max()

The dataset looks like this before running this code:

date	births	year	month	day	lag1	lag2	Roll_mean	Roll_max
1959-01-01	35	1959	1	1	NA	NA	NA	NA
1959-01-02	32	1959	1	2	35	NA	33.5	NA
1959-01-03	30	1959	1	3	32	NA	31.0	35
1959-01-04	31	1959	1	4	30	NA	30.5	32
1959-01-05	44	1959	1	5	31	NA	37.5	44
1959-01-06	29	1959	1	6	44	NA	36.5	44
1959-01-07	45	1959	1	7	29	NA	37.0	45
1959-01-08	43	1959	1	8	45	NA	44.0	45
1959-01-09	38	1959	1	9	43	NA	40.5	45
1959-01-10	27	1959	1	10	38	NA	32.5	43

The dataset looks like this after running this code:

The code he use in python is Feature['Expand_Max']=df['births'].expanding().max()

date	births	year	month	day	lag1	lag2	Roll_mean	Roll_max	Expand_Max
1959-01-01	35	1959	1	1	NA	NA	NA	NA	35
1959-01-02	32	1959	1	2	35	NA	33.5	NA	35
1959-01-03	30	1959	1	3	32	NA	31.0	35	35
1959-01-04	31	1959	1	4	30	NA	30.5	32	35
1959-01-05	44	1959	1	5	31	NA	37.5	44	44
1959-01-06	29	1959	1	6	44	NA	36.5	44	44
1959-01-07	45	1959	1	7	29	NA	37.0	45	45
1959-01-08	43	1959	1	8	45	NA	44.0	45	45
1959-01-09	38	1959	1	9	43	NA	40.5	45	45
1959-01-10	27	1959	1	10	38	NA	32.5	43	45

Solution

You can use zoo::na.locf with fromLast = TRUE which will fill the NA values with the last non-NA value in the column, cummax would return cumulative maximum at every point.

df$Roll_max <- cummax(zoo::na.locf(df$Roll_max, fromLast = TRUE))
df$Roll_max
#[1] 35 35 35 35 44 44 45 45 45 45