The call of your own rf.professionals object suggests united states that the arbitrary tree generated five hundred different woods (the brand new standard) and you may sampled one or two variables at each and every split up. 68 and you may almost 53 percent of the difference told me. Let's see if we can boost to your default amount of woods. So many woods can cause overfitting; needless to say, exactly how many is too of many depends on the knowledge. Some things can help away, the initial one is a storyline from rf.advantages while the most other will be to request minimal MSE: > plot(rf.pros)

This plot shows the brand new MSE of the level of trees when you look at the new design. You can find you to definitely because the woods is actually additional, significant change in MSE takes place early on after which flatlines just just before 100 woods are created regarding the tree. We are able to identify the particular and you will max forest toward hence.min() form, as follows: > which.min(rf.pros$mse) 75

We could is actually 75 woods regarding random tree by simply indicating ntree=75 on design sentence structure: > lay.seed(123) > rf.gurus.2 rf.benefits.dos Telephone call: randomForest(algorithm = lpsa