CODEWITHSUNDEEP
rfor-loopforeach
oercim
oercim

Reputation: 1848

Applying foreach library to a multi regression analysis in R

I have a data frame(df) which has numeric variables. I want to make regression analysis using all combinations of columns(dependent vs. independent variable doesn't matter for me).

So I wrote below code which works fine:

#gtools required for combinations function  
library(gtools)
#generate data-3 columns 30 rows data frame
df<-as.data.frame(replicate(3, rnorm(30))) 
#extract two combination of columns
comb<-combinations(n=ncol(df),2,colnames(df))

listc<-list()
for(i in 1:nrow(comb))
  {
  vars<-df[comb[i,]]
  model.lm<-lm(vars[,1]~vars[,2],data=df)
  listc[[i]]<-coefficients(model.lm)
}

I want to execute this code faster. I try foreachlibrary to enable parallel processing. However, I can't manage to apply the above application?

How can I apply foreach to this code? I will be very glad for any help. Thanks a lot.

Upvotes: 0

Views: 961

Answers (1)

shekeine
shekeine

Reputation: 1465

#Load required libraries
library(parallel)
library(foreach)
library(doParallel)

#Register parallel cluster with all cores available minus 1
cl <- makeCluster(detectCores() - 1)
registerDoParallel(cl, cores = detectCores() - 1)

#Extract two combination of columns
comb <- combinations(n=ncol(df),2,colnames(df))

listc <-  foreach (i=1:nrow(comb), .packages="gtools", .combine='c') %dopar% {

          #Do stuff
          vars <- df[comb[i,]]
          model.lm <- lm(vars[,1]~vars[,2],data=df)

          #Get coeeficients
          coef_i <- list(coefficients(model.lm))
          coef_i

          }
stopCluster(cl)

The returned listc is a list with the coefficients from each iteration / row in df.

Upvotes: 1

Related Questions