如何在循环中使用 which.min(i) 来返回每个变量中每组的最小值？

Question

I´m kind new to R programming and I´m trying to get the minimum value by group in each variable. I have more than 300 variables and am trying to run which.min() inside a loop. Here´sa dummy dataset:

df <- data.frame("Set" = c(rep("A",3),rep("B",3),rep("C",3)),
                 "Pair" = rep(c("a","b","c"),3),
                 "y" = c(4,5,6,4,5,8,9,8,7), 
                 "x" = c(11,13,10,15,14,16,12,18,17),
                 "z" = c(19,20,21,22,23,24,25,26,27))

data:

  Set Pair y  x  z
1   A    a 4 11 19
2   A    b 5 13 20
3   A    c 6 10 21
4   B    a 4 15 22
5   B    b 5 14 23
6   B    c 8 16 24
7   C    a 9 12 25
8   C    b 8 18 26
9   C    c 7 17 27

I´m trying:

library(data.table)
RES <- setDT(df[,c(1,2,3)])[ , .SD[which.min(y)], by = Set]
for (i in 2:3){
  df2 <- as.data.frame(df[,c(1,2,..i+2)])
  res2 <- setDT(df2)[ , .SD[which.min(i+2)], by = Set]
  RES <- cbind(RES,res2)
  rm(res2)
}

My desired output:

  Set Pair y Set.1 Pair.1  x Set.2 Pair.2  z
1   A    a 4     A      c 10     A      a 19
2   B    a 4     B      b 14     B      a 22
3   C    c 7     C      a 12     C      a 25

The problem is within which.min() that does not accept i+2 or nor even i. How do I iterate through columns using which.min()? I tried other functions too, but the closest I got was with this one. I could do just res2 <- setDT(df)[, .SD[which.min(x)], by = Set] , but I have many many columns to go through. If you have another solution, I would be very happy to learn. Thank you!

Answer 1

Here's an approach which uses lapply to loop through the columns of interest of y:z (eg, x, y, and z). If we could simplify and only worry about the minimum, this is what it would look like:

library(data.table)
dt = as.data.table(df)
dt[, lapply(.SD, min), by = Set, .SDcols = y:z]

However, you are interested in both the matching Pair with the minimum value of the columns. To do that, we use lapply to instead return the two values of interest. Then, to have the data combine correctly, we do.call('c', ...)

library(data.table)
dt = as.data.table(df)
dt[, do.call('c', 
             lapply(.SD,
                    function(x) {
                      wm = which.min(x)
                      list(pair = Pair[wm],val =  x[wm])
                      })),
   by = Set,
   .SDcols = y:z]

##    Set y.pair y.val x.pair x.val z.pair z.val
## 1:   A      a     4      c    10      a    19
## 2:   B      a     4      b    14      a    22
## 3:   C      c     7      a    12      a    25