R: finding column with minimum value in each row when there is a tied

Here is my data example:

>dat <- matrix(c(59,50,48,44,44,NA,78,59,42,67,51,NA,开发者_JAVA技巧72,64,64),byrow=TRUE,ncol=3) 
>k <- apply(dat, 1, function(x) which(x == min(x, na.rm = TRUE)))
>k
[[1]]
[1] 3

[[2]]
[1] 1 2

[[3]]
[1] 3

[[4]]
[1] 2

[[5]]
[1] 2 3

But, I want the output like this:

k

3 2 3 2 3

Many thanks in advance.

do you want a maximum index for each row?
then,

> k <- apply(dat, 1, function(x) max(which(x == min(x, na.rm = TRUE))))
> k
[1] 3 2 3 2 3

will do that.

You can use max.col(-dat, "last"), but you'll have to set NAs to Inf first.

You can use this command to apply some functions on multiple (selected) columns of each row. Here I am using this to create a new column for max of columns 1 and 2 (maxv12):

d2<-transform(d, maxv12=apply(d[,c(1,2)],1, max, na.rm = TRUE))

My original data (d) is:

> head(d)
      V1     V2     V3     V4
1 2.0960 3.5364 2.2627 3.4358
2 1.7210 3.3172 1.6559 3.3083
3 1.7950 3.2874 2.2214 3.8520
4 2.0187 3.4038 1.9036 3.4158
5 1.8991 3.6274 1.8083 3.4552
6 1.7382 3.1765 2.6270 4.0960

And applying that command would give me this:

> head(d2)
      V1     V2     V3     V4 maxv12
1 2.0960 3.5364 2.2627 3.4358 3.5364
2 1.7210 3.3172 1.6559 3.3083 3.3172
3 1.7950 3.2874 2.2214 3.8520 3.2874
4 2.0187 3.4038 1.9036 3.4158 3.4038
5 1.8991 3.6274 1.8083 3.4552 3.6274
6 1.7382 3.1765 2.6270 4.0960 3.1765