In jeremyrcoyle/mangolassi: The Scalable Highly Adaptive Lasso

library(knitr)
knitr::opts_chunk$set(echo = TRUE)
library(sl3)
library(delayed)
library(SuperLearner)
library(future)
library(ggplot2)
library(data.table)
library(stringr)
library(scales)

Introduction

This document consists of some simple benchmarks for various choices of SuperLearner implementation, wrapper functions, and parallelization schemes. The purpose of this document is two-fold:

1. Compare the computational performance of these methods
2. Illustrate the use of these different methods

Test Setup

Test System

uname <- system("uname -a", intern = TRUE)
os <- sub(" .*", "", uname)
if(os=="Darwin"){
  cpu_model <- system("sysctl -n machdep.cpu.brand_string", intern = TRUE)
  cpus_physical <- as.numeric(system("sysctl -n hw.physicalcpu", intern = TRUE))
  cpus_logical <- as.numeric(system("sysctl -n hw.logicalcpu", intern = TRUE))
  cpu_clock <- system("sysctl -n hw.cpufrequency_max", intern = TRUE)
  memory <- system("sysctl -n hw.memsize", intern = TRUE)
} else if(os=="Linux"){
  cpu_model <- system("lscpu | grep 'Model name'", intern = TRUE)
  cpu_model <- gsub("Model name:[[:blank:]]*","", cpu_model)
  cpus_logical <- system("lscpu | grep '^CPU(s)'", intern = TRUE)
  cpus_logical <- as.numeric(gsub("^.*:[[:blank:]]*","", cpus_logical))
  tpc <- system("lscpu | grep '^Thread(s) per core'", intern = TRUE)
  tpc <- as.numeric(gsub("^.*:[[:blank:]]*","", tpc))
  cpus_physical <- cpus_logical/tpc
  cpu_clock <- as.numeric(gsub("GHz","",gsub("^.*@","",cpu_model)))*10^9
  memory <- system("cat /proc/meminfo | grep '^MemTotal'", intern = TRUE)
  memory <- as.numeric(gsub("kB","",gsub("^.*:","",memory)))*2^10
} else {
  stop("unsupported OS")
}

• CPU model: r cpu_model
• Physical cores: r as.numeric(cpus_physical)
• Logical cores: r as.numeric(cpus_logical)
• Clock speed: r as.numeric(cpu_clock)/10^9GHz
• Memory: r round(as.numeric(memory)/2^30, 1)GB

Test Data

Tests

microbenchmark({
   glmnet::glmnet(x = x_basis, y = y, intercept = TRUE, nlambda = 100,
   lambda.min.ratio = 0.01, family = "gaussian", alpha = 1, standardize = TRUE)
 }, times = 10)
microbenchmark({
   lassi(x_basis, y, nlambda=100, lambda_min_ratio = 0.01, center = FALSE)
 }, times = 10)

microbenchmark({
   lassi(x_basis, y, nlambda=100, lambda_min_ratio = 0.01, center = TRUE)
 }, times = 10)

microbenchmark({
   glmnet::cv.glmnet(x = x_basis, y = y, intercept = TRUE, nlambda = 100,
   lambda.min.ratio = 0.01, family = "gaussian", alpha = 1, standardize = TRUE)
 }, times = 1)

microbenchmark({
   cv_lasso(x_basis, y, center = FALSE)
 }, times = 1)

microbenchmark({
   cv_lasso(x_basis, y, center = TRUE)
 }, times = 1)

set.seed(1234)
cv_l_full <- cv_lasso(x_basis, y, center = FALSE)
set.seed(1234)
cv_l_es <- cv_lasso_early_stopping(x_basis, y)
plot(cv_l_es)
plot(cv_l_full$lambdas_cvmse)
microbenchmark({
   cv_lasso_early_stopping(x_basis, y)
 }, times = 1)

Session Information

sessionInfo()

jeremyrcoyle/mangolassi documentation built on Nov. 18, 2023, 6:22 p.m.