Function that returns three highest value in row in R

Question

I have two sets of data frames and I am trying to create a function which takes in a data frame and row name as an argument and returns the three highest value on the row (in a descending order) and the name of the column of three highest value.


    set.seed(0)
    df <- data.frame(A=c(3,2,1,4,5),B=c(1,6,3,8,4),C=c(2,1,4,8,9), D=c(4,1,2,4,6))
    row.names(df)<-c("R1","R2","R3","R4","R5")

    df2 <- data.frame(E=c(2,5,6,1,4),F=c(2,4,2,5,1),G=c(5,6,2,7,3),H=c(8,2,7,4,1))
    row.names(df2)<-c("R6","R7","R8","R9","R10")

    print(df)

       A B C D
    R1 3 1 2 4
    R2 2 6 1 1
    R3 1 3 4 2
    R4 4 8 8 4
    R5 5 4 9 6

    print(df2)

        E F G H
    R6  2 2 5 8
    R7  5 4 6 2
    R8  6 2 2 7
    R9  1 5 7 4
    R10 4 1 3 1

Here is an example of a result:

Let the function be maxthree. Now


    maxthree(df2, "R7")

    G E F
    6 5 4

Here is what I have done so far:


    maxthree <- function(data,row) {
      if(!row %in% rownames(data)) {
        print("Check value")
      } else { 
        max_col <- which.max(data[row,])
        print(max_col)
      }
    }

This function will now return the maximum value in that row as well as the column name. However, I don't now how to add the second and the third highest values to the function.

What happens if you want both R6 and R7 for example? – NelsonGon May 12 '20 at 16:48 — NelsonGon, May 12 '20 at 16:48

score 3 · Accepted Answer · answered May 12 '20 at 16:40

3

maxthree = function(data, row) {
  data[row, order(unlist(data[row, ]), decreasing = TRUE)[1:3]]
}

maxthree(df2, "R7")
#    G E F
# R7 6 5 4

The result is a 1x3 data frame.

answered May 12 '20 at 16:40

Gregor Thomas

136,190
20
167
294

score 3 · Answer 2 · answered May 12 '20 at 16:40

3

This should work great

maxthree <- function(data,roww){
    x <- data[roww,] 
    x[order(x, decreasing = T)][1:3]
}


>  maxthree(df2, "R7")
   G E F
R7 6 5 4

answered May 12 '20 at 16:40

Daniel O

4,258
6
20

score 2 · Answer 3 · answered May 12 '20 at 16:39

Try this:

df <- data.frame(A=c(3,2,1,4,5),B=c(1,6,3,8,4),C=c(2,1,4,8,9), D=c(4,1,2,4,6))
row.names(df)<-c("R1","R2","R3","R4","R5")

df2 <- data.frame(E=c(2,5,6,1,4),F=c(2,4,2,5,1),G=c(5,6,2,7,3),H=c(8,2,7,4,1))
row.names(df2)<-c("R6","R7","R8","R9","R10")

maxthree <- function(data,row) {
  named_vec <- t(data)[,row]
  return(sort(named_vec, decreasing = T)[1:3])
}

maxthree(df2, "R7")

# G E F 
# 6 5 4

This approach transposes your data frame "t()" to allow a straightforward subset of the row as a named vector. This allows sort to be used to order the values as desired.

Transposing the entire data frame is unnecessary and will slow this down on even moderately sized data--extract the row first and then transpose: `t(data[row, ])` instead of `t(data)[, row]` for a much more efficient solution. — Gregor Thomas, May 12 '20 at 16:53

GKi · Answer 4 · 2020-05-13T08:09:44.083

2

You can use sort and [1:3] to get the first 3 elements like:

maxthree <- function(data,row) {sort(data[row,], TRUE)[1:3]}
maxthree(df2, "R7")
#   G E F
#R7 6 5 4

In case the rowname should not be shown you can add unlist:

maxthree <- function(data,row) {head(unlist(sort(data[row,], TRUE)),3)}
maxthree(df2, "R7")
#G E F 
#6 5 4

edited May 13 '20 at 08:09

answered May 12 '20 at 16:59

GKi

37,245
2
26
48

1

Interesting! I didn't expect `sort` to work on a data frame, but when it's a single row it does. – Gregor Thomas May 13 '20 at 13:54

score 1 · Answer 5 · answered May 12 '20 at 16:45

1

You can use the order function.

maxthree <- function(data, row_name) data[row_name, order(-data[row_name,])][, 1:3]
maxthree(df2, 'R7')
   G E F
R7 6 5 4

answered May 12 '20 at 16:45

user2474226

1,472
1
9
9

Using `-data[row_name, ]` instead of the `decreasing = TRUE` argument limits your function to only work with numeric data. `decreasing = TRUE` makes it a little more flexible just in case it's ever needed for, say, `character` data. – Gregor Thomas May 12 '20 at 16:56

Function that returns three highest value in row in R

5 Answers5