Add rollapply vector to a data frame

Question

I have the following data:

Year Month Total
2015 1     123
2015 2     435
2015 3     543

which are total by month over a three year period. I used rollapply to calculate a rolling cumulative total. Like below:

rollapply(results$Total, 12, sum)

however this returns a vector which (obviously) has 11 fewer observations and no index value.

How do I join it back to the dataframe as a column indexed correctly?

Answer 1

You can add the vector to your main dataframe as a new column:

results$rolling <- rollapply(results$Total, width = 3, sum , fill = NA)

then you should get the results as:

Year Month Total rolling
2015 1     123   na
2015 2     435   na
2015 3     543   1101

I changed the window size to 3 just for illustration in this example!

Answer 2

Use library zoo. val is for the column where you want to get the sum

library(zoo)

rsum <- function(df, val, k) {
  df %>%
    mutate(
      rsum = rollapply(val, k, FUN = sum, partial=T, align="center"))
    )  
}

df %>% rsum(., .$col, 12)  

if there's groupping information, then

df %>% group_by(a,b,c) %>% group_modify(~ { .x %>% rsum(.x$col,12)})