aggregating the data in different columns based on condition in r - r

Question

order_id customer_id extension_of quantity cost duration
1 123 srujan NA 1 100 30
2 456 teja NA 1 100 30
3 789 srujan 123 1 100 30
I have sample data which contains order information. What I need to do is summarize the data(sum of cost etc) if value of order_id and extension_of columns matches.

Answer 1

I take it you want to add rows #1 and #3 as #1 is like the parent order and #3 is its extension. Here's one way you can do it (assuming dft is your data frame):
library(purrr)
dft$parent_id <- map2_dbl(dft$order_id, dft$extension_of, function(order_id, extension_of) if(is.na(extension_of)) order_id else extension_of)
aggregate(cost~parent_id, data=dft, FUN=sum)
This can not do recursion (i.e. an extension being extended itself), but it does sum up for one level.
If you need several levels, you can go for something like this:
library(purrr)
find_root <- function(order_id, extension_of){
if(is.na(extension_of)){
return(order_id)
}else{
parent_ext <- dft$extension_of[dft$order_id==extension_of]
return(find_root(extension_of, parent_ext))
}
}
dft$parent_id <- map2_dbl(dft$order_id, dft$extension_of, find_root)
aggregate(cost~parent_id, data=dft, FUN=sum) # same as before
(not it looks up the parent's parent, if necessary). One can certainly optimize for performance if really needed; the principle will stay the same, however.