Handling ties in finding index of n th maximum value in R - r

Question

I am working on a dataframe and trying to find the index of nth maximum value (n varies by a loop), however, in the columns I have tied values and the program throws an error. Below is a sample dataset. I am basically trying to generate a similar dataframe, but with only the index values of all the values in the column vector of the dataframe.
For the output DF, column 1 in the output DF will have index values of elements of Refer_1, so Output_DF[1,1] will have the index for highest value, while Output_DF[10,1] will have the index of lowest value. Below is the input DF.
Input
1 17
2 21
3 13
4 26
5 204
6 36
7 14
8 25
9 45
10 37
Output (index values)
5
9
10
6
4
8
2
1
7
3
I am currently using which, unlist and partial together to get the indexes, however, I am unable to rectify the error. Note that the ties can occur with any nth maximum value (not necessarily the column maxima).
which(Consolidated_data_new[,i]==unlist(sort(Consolidated_data_new[,i],partial=j)[j]))
Please note that I want the code to return only one value at a time, and handle the 2nd tied value in the next loop iteration.
Please help solve this.
Regards,

Answer 1

library(data.table)
DT<-structure(list(Refer_1 = c(11L, 15L, 7L, 19L, 104L, 24L, 11L,
22L, 39L, 19L), Refer_2 = c(17L, 21L, 13L, 25L, 204L, 36L, 14L,
25L, 45L, 37L)), .Names = c("Refer_1", "Refer_2"), row.names = c(NA,
-10L), class = c("data.table", "data.frame"), .internal.selfref = <pointer: 0x0000000000130788>)
DT[,lapply(.SD, order,decreasing=TRUE)]
Refer_1 Refer_2
1: 5 5
2: 9 9
3: 6 10
4: 8 6
5: 4 4
6: 10 8
7: 2 2
8: 1 1
9: 7 7
10: 3 3

Answer 2

Your comments suggest you are working with a dataframe that has more than one column and that you want an output dataframe that has the results of order with decreasing=TRUE applied to every column:
> DF[2] <- sample(1:300, 10)
> DF[3] <- sample(1:300, 10)
> DF
Input V2 V3
1 17 210 3
2 21 72 4
3 13 263 1
4 26 249 6
5 204 223 10
6 36 83 7
7 14 107 2
8 25 295 5
9 45 198 9
10 37 112 8
> ordDF <- as.data.frame(lapply(DF, order, decreasing=TRUE))
> names(ordDF) <- paste0("res", 1:length(DF) )
> ordDF
res1 res2 res3
1 5 8 4
2 9 3 9
3 10 4 2
4 6 5 7
5 4 1 10
6 8 9 8
7 2 10 1
8 1 7 6
9 7 6 3
10 3 2 5
> dput(ordDF)
structure(list(res1 = c(5L, 9L, 10L, 6L, 4L, 8L, 2L, 1L, 7L,
3L), res2 = c(8L, 3L, 4L, 5L, 1L, 9L, 10L, 7L, 6L, 2L), res3 = c(4L,
9L, 2L, 7L, 10L, 8L, 1L, 6L, 3L, 5L)), .Names = c("res1", "res2",
"res3"), row.names = c(NA, -10L), class = "data.frame")