check boolean expression in dataframe Rcpp (C++) - c++

Question

I have a dataframe dat with data and a vector rule with logical rules
set.seed(124)
ro <- round(runif(n = 30,1,10),2)
dat <- as.data.frame(matrix(data =ro,ncol = 3)) ; colnames(dat) <- paste0("x" ,1:ncol(dat))
rule <- c("x1 > 5 & x2/2 > 2" , "x1 > x2*2" , "x3!=4")
I need to check if the expression is true
id <- 2
for(i in 1:nrow(dat)){
cr <- with(data = dat[i,] , expr = eval(parse(text = rule[id])))
print(cr)
}
[1] FALSE
[1] FALSE
[1] FALSE
[1] FALSE
[1] FALSE
[1] TRUE
[1] FALSE
[1] FALSE
[1] FALSE
[1] TRUE
How to do this with Rcpp ?

Answer 1

Two things worth stressing here are
you do not need a low over all rows as R is vectorized, and that already fast
you can sweep the rules over your data and return a result matrix
Both of those are a one-liner:
> res <- do.call(cbind, lapply(rule, \(r) with(dat, eval(parse(text=r)))))
> res
[,1] [,2] [,3]
[1,] FALSE FALSE TRUE
[2,] FALSE FALSE TRUE
[3,] TRUE FALSE TRUE
[4,] FALSE FALSE TRUE
[5,] FALSE FALSE TRUE
[6,] FALSE TRUE TRUE
[7,] TRUE FALSE TRUE
[8,] TRUE FALSE TRUE
[9,] TRUE FALSE TRUE
[10,] FALSE TRUE TRUE
>
(I used the R 4.1.* anonymous function there, you can relace \(r) with the standard function(r) as well.)
As this is already vectorised it will be faster than your per-row call, and even if you did it with Rcpp if would not be (much) faster than already vectorised code.