Question

I wish to write a function using dlply to fit a linear regression stratified by

"cat1 by arg1"

So my function looks like this

fun1 <- function(arg1) {
     m1 <- data.frame(...) 
     mod.var <- ...
     mod.form <- formula(paste("y ~", paste(mod.var, collapse = " + ")))
     list_of_models <- dlply(m1, .(cat1,arg1), function(X) lm(mod.form, data = X, 
          na.action=na.omit), .parallel=FALSE) 
}

How to write a function so that when I call the function fun1("cat2") the function will execute

list_of_models <- dlply(m1, .(cat1,cat2), function(X) lm(mod.form, data = X,
    na.action = na.omit), .parallel=FALSE)

and call the function fun1("cat3"), the function will execute

list_of_models <- dlply(m1,.(cat1,cat3), function(X) lm(mod.form, data = X,
    na.action=na.omit), .parallel=FALSE)

where cat1, cat2 and cat3 are the names of categorical variables.

Thank you for your help.

Edit: As the function stands at the moment, it does not work properly because ".(cat1,arg1)" is not “recognizable” in dlply. Some modifications are needed, but how?

Answer 1

Try c("cat1", arg1) instead of .(cat1, arg1).

Citing ?dlply:

Arguments

.variables variables to split data frame by, as quoted variables, a formula or character vector