sorting data to provide top entries - r

Question

I have a table like below. Each row has store id, discount % for one of their coupons. Each store could have multiple coupons but (store+discount %) is a primary key. I would like to find out top 10 coupons (by decreasing order of discount %) but would like to get only 2 coupon from the same store. What is the most efficient way to do this? My logic involves sorting data multiple times. Is there a better and more efficient way? I would like to do this in R.
Sample data:
df <- data.frame(Store=c("Lowes","Lowes","Lowes","Lowes","HD","HD","HD","ACE",
"ACE","Misc","Misc","Other","Other","Last","Last","Last"),
`discount_%`=c("60%","50%","40%","30%","60%","50%","40%","30%",
"20%","50%","30%","20%","10%","10%","5%","3%"),
check.names = FALSE)
my solution is ignore the store and sort the table by discount then
create a ID. ID would represent coupons in descending order
Then by Store and discount create ID2 which would have rankings of
coupons by store.
then filter all rows where ID2>2
then sort table by ID
take top 10 rows

Answer 1

Try this:
df$`discount_%` <- as.numeric(gsub("%","",df$`discount_%`))
require(data.table)
setDT(df)[order(-`discount_%`),.SD[1:2],by=Store][order(-`discount_%`)[1:10],]
Output:
Store discount_%
1: Lowes 60
2: HD 60
3: Lowes 50
4: HD 50
5: Misc 50
6: Misc 30
7: ACE 30
8: ACE 20
9: Other 20
10: Other 10
Data is easier to work with in R without special characters, but if you need to add the percent sign back, try something like this:
paste0(df$`discount_%`,"%")