[Rd] small performance enhancement suggestion (PR#10409)

[simon . debernard]

Full_Name: Simon de Bernard
Version: 
OS: 
Submission from: (NULL) (140.77.34.213)


In src/main/dotcode.c, function RObjToCPtr, test for naok should be transferred
outside of the loops...

__
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel

[Rd] "by" speed improvement (PR#11064)

[simon . debernard]

Full_Name: Simon de Bernard
Version: 2.7.0 (44733)
OS: MacOS
Submission from: (NULL) (140.77.34.213)


"by" usually takes forever even on a "not so large" data structure.

If one can do with a matrix instead of a data.frame, defining by.matrix as
by.data.frame modified to convert data back to a data.frame in the "eval" call
improves speed by orders of magnitude (whatever that means ;-) )

I'd suggest defining the by.matrix function directly in R...

__
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel

[Rd] (PR#11064) how to reproduce...

[simon . debernard]

You can try this:

data <- cbind("a"=sample(1:10), "b"=sample(1:10))
fact <- sample(rep(1:1, each=10))
system.time(std <- by(data, fact, colSums))
by.matrix <- function (data, INDICES, FUN, ...) {
 if (!is.list(INDICES)) {
 IND <- vector("list", 1)
 IND[[1]] <- INDICES
 names(IND) <- deparse(substitute(INDICES))[1]
 }
 else IND <- INDICES
 FUNx <- function(x) FUN(data[x, , drop = FALSE], ...)
 nd <- nrow(data)
 ans <- eval(substitute(tapply(1:nd, IND, FUNx)), as.data.frame 
(data))
 attr(ans, "call") <- match.call()
 class(ans) <- "by"
 ans
}
system.time(mod <- by(data, fact, colSums))
all.equal(std, mod)

I get a 30x speed up
(I'm not sure why the attributes differ, but I'm sure this can be  
fixed...)

__
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel