Convert numeric variable into binary columns

Question

Suppose I have the following data table in R:

DT <- data.table::data.table(y=runif(1e4), x1=rnorm(1e4), x2=as.factor(sample(1:11,1e4,TRUE)))

Since x2 takes value 1 to 11, the maximum binary representation for 11 is 1011, so 4 columns would be enough, I want to convert x2 to 4 binary columns such that:

y           x1        x2   b1  b2  b3  b4
0.17438022  0.1925023  11  1    0  1   1
0.34850700  1.0412363  3   0    0  1   1

How to do it in R?

This was answered here: https://stackoverflow.com/questions/12088080/how-to-convert-integer-number-into-binary-vector — HolgerBarlt, Oct 16 '18 at 14:50
I'd probably use `digitsBase()` from the **sfsmisc** package. Try, for example, `t(digitsBase(1:11, 2, ndigits=4))`. — Josh O'Brien, Oct 16 '18 at 14:56

Andre Elrico · Answer 1 · 2018-10-16T15:17:06.307

For the converting part I've used THIS Answer.
As already provided by you. DT must be a data.table.

binarys <-
sapply(
    as.numeric(as.character(DT$x2)), function(x) {
        sub(".*(?=.{4}$)", "",
        paste(rev(as.integer(intToBits(x))), collapse=""),
        perl = T)
    })

DT[,c(c(DT), tstrsplit(binarys,""))]

#            y         x1 x2 V4 V5 V6 V7
#1: 0.09963794  0.2799082  8  1  0  0  0
#2: 0.04547423 -0.8783603 11  1  0  1  1
#3: 0.41489062 -0.8319349 10  1  0  1  0
#4: 0.96606376  0.6323427  3  0  0  1  1
#5: 0.50516936 -0.5751176 11  1  0  1  1

sub(".*(?=.{4}$)" ... is used to remove everything except for the last 4 digits. If your numbers get bigger you might want to tweak this.

Convert numeric variable into binary columns

1 Answers1