R: how do I dichotomise efficiently

Question

Is there a more "R-minded" way to dichotomise efficiently? Thanks.

y<-c(0,3,2,1,0,0,2,5,0,1,0,0);b<-vector()

for (k in 1:length(y)) {
    if (y[k] == 0) b[k] = 0
    else
        b[k] = 1
}
y;b

Answer 1

You have something that works. Are you worried about speed for some reason? Here's an alternative:

y<-c(0,3,2,1,0,0,2,5,0,1,0,0)

decider = function( x ) {
   if ( x == 0 ) {
      return(0)
   }

   return(1)
}

b = sapply( y, decider )

Answer 2

Try this:

b <- rep(0, length(y))
b[y != 0] <- 1

This is efficient because y and b are the same size and rep() is very fast/vectorized.

Edit:Here's another approach:

b <- ifelse(y == 0, 0, 1)

The ifelse() function is also vectorized.

Answer 3

b <- as.numeric(y!=0)

Answer 4

Use ifelse(). This is vectorized and (edit: somewhat) fast.

> y <- c(0,3,2,1,0,0,2,5,0,1,0,0)
> b <- ifelse(y == 0, 0, 1)
 [1] 0 1 1 1 0 0 1 1 0 1 0 0

Edit 2: This approach is less fast than the as.numeric(y!=0) approach.

> t <- Sys.time(); b <- as.numeric(y!=0); Sys.time() - t # Rob's approach
Time difference of 0.0002379417 secs
> t <- Sys.time(); b <- ifelse(y==0, 0, 1); Sys.time() - t # Shane's 2nd and my approach
Time difference of 0.000428915 secs
> t <- Sys.time(); b = sapply( y, decider ); Sys.time() - t # James's approach
Time difference of 0.0004429817 sec

But to some, ifelse may be trivially more readable than the as.numeric approach.

Note the OP's version took 0.0004558563 to run.

Answer 5

b<-(y!=0)+0

b [1] 0 1 1 1 0 0 1 1 0 1 0 0