[SOLVED] Apply rolling function to one column using zoo::rollapply in dplyr::mutate

Apply rolling function to one column using zoo::rollapply in dplyr::mutate

I want to apply a rolling function to a vector in a data frame, specifically, to take the difference between a single column for every 10 observations. For example, to get the difference in Sepal.Lengths from the iris dataset, I tried the following:

data <- data(iris)

data1 <- data %>%
  mutate(length_Change = zoo::rollapply(Sepal.Length, 
                                   width = 10, 
                                   FUN = function(x){max(x, na.rm = T)-min(x, na.rm = T)}, 
                                   align = "right", 
                                   by.column = FALSE))

but I got the following error:

Error in `mutate()`:
ℹ In argument: `length_Change = rollapply(...)`.
Caused by error:
! `length_Change` must be size 150 or 1, not 141.

and I can't figure out why this isn't working. Any ideas?

Thanks!

Solution

The issue is that the size of your window is 10, the number of rows in iris is 150, and (by default) partial = FALSE, so once it hits the 141st row, and there's only 9 available rows remaining in the dataframe, so it stops there (i.e. it doesn't run on a partial window), hence you getting 141 rows.

To get it working, set partial = TRUE.

For more information, read the documentation.