CODEWITHSUNDEEP
rrandom-forestcbind
DGenchev
DGenchev

Reputation: 367

Aggregate (RF) Variable Importance for Many Forests

I have been testing random forests on one dataset. I would like to output the variable importance of all RFs in one data frame. Something like:

forests <- grep("rf", ls(), value=T)
importances <- do.call(cbind, lapply(forests, importance))

which throws an error:

Error in UseMethod("importance") : no applicable method for 'importance' applied to an object of class "character"

I tried converting forests to a list, but that didn't help either.

Example:

rf10 <- randomForest(mpg ~., mtcars, ntree=10)
rf100 <- randomForest(mpg ~., mtcars, ntree=100)
cbind(importance(rf10), importance(rf100))

Upvotes: 0

Views: 266

Answers (1)

jav
jav

Reputation: 1495

You should do instead

do.call(cbind, lapply(forests, function(x) importance(get(x))))

The return value from grep was a list of variable names, not the variables themselves. When you did importance(x), it was for example doing importance("rf10"). You should be using the object as the parameter, and not the object's name. get(x) returns the reference object for you.

Upvotes: 1

Related Questions