Loops with if statements classifying numbers

Question

I am having trouble with a loop I am creating. My intention is that the loop will see where a value falls and give that value another name. Here is an example of what I am trying to do:

a<-rnorm(10,0,1)
b<-rnorm(10,0,1)

testing<-data.frame(a,b)

testing2<-testing
for (i in 1:nrow(testing2)){
  for (j in 1:ncol(testing2)){
    if (testing2[i,j]>1) testing2[i,j]<-"More"
    else if (testing2[i,j]<(-1)) testing2[i,j]<-"Less"
    else testing2[i,j]<-"Same"
  }
}

When I look at testing2 and compare it to testing, it does not match what it should be doing. "More" seems to work, but it gets mixed up between "Less" and "Same".

score 5 · Accepted Answer · answered Mar 04 '14 at 16:13

5

No need for a (nested) loop, use the vectorised ifelse instead:

testing = ifelse(testing > 1, 'More', ifelse(testing < -1, 'Less', 'Same'))

answered Mar 04 '14 at 16:13

Konrad Rudolph

530,221
131
937
1,214

A5C1D2H2I1M1N2O1R2T1 · Answer 2 · 2014-03-04T16:39:21.103

5

cut comes to mind as an obvious candidate here. Since you need to apply it across multiple columns, you can use lapply as follows:

x <- testing
x[] <- lapply(testing, cut, breaks=c(-Inf, -1, 1, Inf), 
              labels=c("less", "same", "more"))
x

^^edited with refinements from the comments ;-)

Or with as.matrix instead of lapply...

x <- cut(as.matrix(testing), c(-Inf, -1, 1, Inf), 
         labels=c("less", "same", "more"))
dim(x) <- dim(testing)

edited Mar 04 '14 at 16:39

answered Mar 04 '14 at 16:24

A5C1D2H2I1M1N2O1R2T1

190,393
28
405
485

1

+1 beat me to it, although I was going to go with the more compact `lapply(testing, cut, breaks=c(-Inf, -1, 1, Inf), labels=c("less", "same", "more"))` – Matthew Plourde Mar 04 '14 at 16:27
@MatthewPlourde, I need to break my habit of unnecessary anonymous functions :-) – A5C1D2H2I1M1N2O1R2T1 Mar 04 '14 at 16:28
1

Worse habits exist. I like your use of `x[]`, this never occurred to me. – Matthew Plourde Mar 04 '14 at 16:32
`cut` was the function I was missing, and hence ended up with the (horrible) nested `ifelse`. Very nice. An addition to Matthew’s remark about the unnecessary lambda I just wanted to add that the assignment (`A <- …`) is also unnecessary. – Konrad Rudolph Mar 04 '14 at 16:36
@KonradRudolph, ha! That was a relic of me originally thinking about if `factor` would work before I remembered `cut`. Oops. – A5C1D2H2I1M1N2O1R2T1 Mar 04 '14 at 16:37

Nishanth · Answer 3 · 2014-03-04T16:45:52.720

1

As suggested, there are better ways of doing this. But coming back to your code, you are messing things up by assigning a character to numeric type.

In R data.frame the columns have the same type. Therefore once you assign a character, the entire column type becomes character.

You can verify this by calling sapply(testing2, class) before and after the for loops

edited Mar 04 '14 at 16:45

answered Mar 04 '14 at 16:39

Nishanth

6,932
5
26
38

Loops with if statements classifying numbers

3 Answers3

Linked

Related