inst/benchmarks/logistic.r

In wrathematics/linmod: Linear and Generalized Linear Model Fitters

library(memuse)
library(linmod)
library(microbenchmark)



burnin <- function(reps=10)
{
  x <- matrix(rnorm(30), 10)
  y <- rnorm(10)
  
  #replicate(RcppEigen::fastLm(X=x, y=y), n=reps)
  replicate(lm_fit(x=x, y=y), n=reps)
  replicate(lm.fit(x=x, y=y), n=reps)
  
  invisible()
}
burnin()


reps <- 15

m <- 5000
n <- 250

x <- matrix(rnorm(m*n), m, n)
y <- rnorm(m)



cat("------------------ RRQR ------------------\n")
cat(paste0("Data size:  ", object.size(x)+object.size(y), "\n"))
cat(paste0("L3 cache size:  ", Sys.cachesize()$L3, "\n"))



control <- list(warmup=20, order="random")

bench <- microbenchmark(
          lm.fit(x=x, y=y),
          lm_fit(x=x, y=y, check.rank=TRUE),
#          RcppEigen::fastLm(X=x,  y=y, method=0), 
          times=reps,
          unit="s",
          control=control
)

print(bench)

n <- 5000
p <- 300

family <- binomial(logit)

linmod:::glm_test(n, p, family, verbose=TRUE, timings=TRUE)