Transform many numerical variables to dummy variables at once

Question

I'm stuck trying to convert a bunch of integer variables (with values from one to 10, plus a few NAs) into the same number of boolean variables, where values<=8 become FALSE, >8 become TRUE, and NAs remain as NAs.

Based on other Q&As around, I tried the following but nothing worked:

TPS <- names(datos[,1:3])
    for(i in TPS) {
     data4[[i]]==NA <- NA
     data4[[i]]<8 <- FALSE
     data4[[i]]>=8 <- TRUE
    }

    TPS <- names(data4[,8:48])
    for(i in TPS) {
      data4[,i]==NA <- NA
      data4[,i]<8 <- FALSE
      data4[,i]>=8 <- TRUE
    }

    TPS <- names(data4[,8:48])
    for(i in TPS) {
      ifelse(data4[,i]<8, FALSE, TRUE)
    }

I know how to do it one variable at a time, but my dataset have 64 variables I have to transform in just 2 different ways, so a procedure to do this in just 2 steps would be extremely useful.

Just tried "data4[,8:48]<8<-FALSE based on akrun's suggestion, but it returns an error message: "incorrect number of arguments to "<<-" — Ignacio1981, Aug 05 '15 at 16:30
If it is a normal dataset with numeric columns, it should work. My code was just `data4[,8:48] >8` and not `data4[,8:48] <8 <- FALSE`. The output would be a logical matrix of TRUE, FALSE, and NA. So, you don't need to assign to FALSE again. — akrun, Aug 05 '15 at 16:32

score 3 · Accepted Answer · answered Aug 05 '15 at 16:44

3

You can just use

 data4[,8:48] >8

The output will be a logical matrix of TRUE/FALSE values and NA (if there are).

answered Aug 05 '15 at 16:44

akrun

874,273
37
540
662

Transform many numerical variables to dummy variables at once

1 Answers1