data.table 评估 1/0 不等于 TRUE/FALSE

Question

Consider the code snippet:

library(data.table)
foo <- data.table(a = rep(c(1, 0), 5), b = c(rep('bar', 5), rep('baz', 5)))

then this is correct:

> foo[, sum(b == 'bar')]
[1] 5

But this is not:

> foo[, sum(b[a] == 'bar')]
[1] 5

Which could be corrected by:

> foo[, sum(b[a == 1] == 'bar')]
[1] 3

Is 1/0 is not evaluated as TRUE/FALSE in this case?

Answer 1

This is a case where printing out intermediate results can help out a lot with what's going on.

a = rep(c(1, 0), 5) 
b = c(rep('bar', 5), rep('baz', 5))
a
#>  [1] 1 0 1 0 1 0 1 0 1 0
b
#>  [1] "bar" "bar" "bar" "bar" "bar" "baz" "baz" "baz" "baz" "baz"
b[a]
#> [1] "bar" "bar" "bar" "bar" "bar"
b[as.logical(a)]
#> [1] "bar" "bar" "bar" "baz" "baz"

^{Created on 2019-10-28 by the reprex package (v0.3.0)}

By using b[a] what you're telling R is that for each element of a you want the element of b that corresponds to the index that comes from a . And since a consists of only zeros and ones, you end up just getting the first index 5 times.