How to select certain rows of a data set in R to then use in a function?

Question

I am trying to find the Mahalanobis Distance between the different species in the iris dataset in R. I was able to find the distance between setosa and versicolor by the following code:

library(HDMD)

#To get Mahalanobis distances between Setosa and Versicolor,
set.vers<-pairwise.mahalanobis(x=iris[1:100,1:4], grouping=iris[1:100,]$Species)
md= sqrt(set.vers$distance)

However, I am struggling to do the same for setosa and virginica. I am not sure how to select the first 50 rows and last 50 rows of the data set (i.e. not have any versicolor data)

score 2 · Answer 1 · answered Sep 24 '20 at 09:40

2

This is a basic subsetting question. You want to subset based on Species, something along the lines of (not tested)

ss <- iris[iris$Species %in% c("Setosa", "Virginica"), ]
pairwise.mahalanobis(x = ss, grouping = ss$Species)

You can of course change the species pair you want to compare in many ways.

answered Sep 24 '20 at 09:40

Roman Luštrik

69,533
24
154
197

When I use that, is says that ss has 0 rows. the following error appears '[1] Sepal.Length Sepal.Width Petal.Length Petal.Width Species <0 rows> (or 0-length row.names)' – daisybeats Sep 24 '20 at 10:14
@daisybeats The error comes from upper cases in `"Setosa"` and `"Virginica"`. Once corrected, in the second line it must be `x = ss[-5]` or `x = ss[1:4]`. And `grouping = droplevels(ss$Species)` would avoid the warning message. – Rui Barradas Sep 24 '20 at 15:08
1

@RuiBarradas you're right, calculating the distance on non-numeric Species column is a bit harder to do. – Roman Luštrik Sep 26 '20 at 16:25

score 1 · Accepted Answer · answered Sep 24 '20 at 10:27

Here is a way to get all combinations of levels in iris$Species with combn and compute the Mahalanobis distances.

library(HDMD)

inx <- sapply(levels(iris$Species), function(l) which(iris$Species == l), simplify = FALSE)
inx <- combn(inx, 2, function(x) unlist(x), simplify = FALSE)
set.vers_all <- lapply(inx, function(i) {
  pairwise.mahalanobis(x = iris[i, 1:4], grouping = droplevels(iris$Species[i]))
})
set.vers_all

How to select certain rows of a data set in R to then use in a function?

2 Answers2