Gini Index Random Forest

TimboTimbo Member Posts: 14 Contributor II
edited November 2018 in Help
Hi,

whenever I try using the gini index option for the Random Forest the process breaks down due to memory issues. Even if I only use something like 7000 examples, 15 Attrubutes and 2 classes for which all other options work fine. Is that a known bug or just due to the method?

Answers

  • landland RapidMiner Certified Analyst, RapidMiner Certified Expert, Member Posts: 2,531 Unicorn
    Hi,
    please restrict the maximal tree depth and minimal number of examples per leaf to cut tree building. Might be it "explodes" exponentially.

    Greetings,
      Sebastian
  • TimboTimbo Member Posts: 14 Contributor II
    Well I am using default settings. if it is "exploding" with those then maybe this shouldn't be default settings...
  • landland RapidMiner Certified Analyst, RapidMiner Certified Expert, Member Posts: 2,531 Unicorn
    Hi,
    if it can explode depends on your data: If you have many nominal values, the tree size can be much greater than if all attributes are numerical and there can be only a split into two branches...

    Greetings,
      Sebastian
Sign In or Register to comment.