Pfthb

I would like to use the shift function from data.table to lead/lag a new column, but I would like to recycle values from the lagged vector that was added to the data.table. From what I can see, fill must be a vector of length 1, and so the values that are lagged must be populated with a constant value (ie NA here).

Please see the MWE below.

dt1 is the resulting data.table using the shift function as is. The new b column has NA values where 4, 5, and 6 should be.

dt2 is the desired data.table result. If my thinking is correct, the output requires R recycling rules but with a lead/lag value specified where the vector should begin.

I could have added a new vector (see below in x_to_avoid) but that requires more manual work that I hope to avoid.

Thanks,

library(data.table)
library(magrittr)

# vector to lead/lag when updating datatable
x = c(1:6)

# leaves NA where 4, 5, 6 "should" have gone for my purposes
dt1 <- data.table(a = c(1:10)) %>% 
 .[, b := shift(x,
 n = 3L,
 fill = NA,
 type = c("lag"))]
dt1

# desired output
dt2 <- dt1[, .(a)] %>% 
 .[, b := c(4,5,6,1,2,3,4,5,6,1)]

# could use another vector, but my actual use is more complicated and I prefer to avoid this (if possible)
x_to_avoid = c(4,5,6,1:6,1)

I think that binhf::shift does the job. First you will need to lengthen the vector using rep.len and then you can cycle it using binhf::shift. I have no idea about the performance though.