CODEWITHSUNDEEP
rsequence
Nakx
Nakx

Reputation: 1600

Finding "which number is not" in a numeric sequence in R

When trying to exclude specific numbers from a sequence I have noticed that which(!(0:10 %in% 2:3)) returns a different result (removes 3 and 4 and add 11) than setdiff(0:10,2:3) or which(!(1:10 %in% 2:3)).

which(!(1:10 %in% 2:3))
[1]  1  4  5  6  7  8  9 10
which(!(0:10 %in% 2:3))
[1]  1  2  5  6  7  8  9 10 11
setdiff(0:10,2:3)
[1]  0  1  4  5  6  7  8  9 10

This seems to be a simple logic problem but I can't figure what is this due to? Is setdiff as fast as which(!()) for large sequences?

Upvotes: 1

Views: 544

Answers (1)

Nakx
Nakx

Reputation: 1600

Summarizing the comments, which gives the position of the elements, not the elements themselves, the example above is unclear but these examples make it obvious:

which(!(10:20 %in% 12:13))
[1]  1  2  5  6  7  8  9 10 11

0:10 %in% 2:3
[1] FALSE FALSE  TRUE  TRUE FALSE FALSE FALSE FALSE FALSE FALSE FALSE

So the correct way of selecting the elements would be:

myseq<-0:10
myseq[which(!(0:10 %in% 2:3))]

Regarding speed, selecting the elements directly from the vector is slightly faster:

myseq<-0:1000000
> microbenchmark::microbenchmark(
+   wh=myseq[which(!(0:1000000 %in% 2:3))],
+   sd=setdiff(0:1000000,2:3),
+   seq=(0:1000000)[!0:1000000 %in% 2:3]
+ )
Unit: milliseconds
 expr      min       lq     mean   median       uq      max neval cld
   wh 18.62157 18.85489 25.17644 25.37830 26.89162 152.9487   100  a 
   sd 36.09655 42.83383 50.22088 44.16595 45.96227 178.2949   100   b
  seq 17.51332 17.98346 25.00993 24.39265 25.91137 174.1875   100  a

Upvotes: 1

Related Questions