Suppose that I am using a continuous variable as an independent variable (although a bad choice) for C 4.5 or C 5.0 (tree based classifier). I am having a hard time figuring out does C 4.5 or C 5.0 does a binary split or a multi way split by calculating entropy?

If it does a multi way split how does it decides on optimal number of splits?

If it does a binary split for continuous variable using Entropy or Gini Index, then it's alright to regard it's splitting criteria the same as of CART?

  • $\begingroup$ All the implementations that I know of use only binary splits. Multi way splits are decided beforehand, and there are always the same amout of them at every node (unless the methods reaches some stopping criteria first). $\endgroup$ Commented Oct 11, 2018 at 8:33
  • $\begingroup$ I think you'd be more likely to get an answer in crossvalidated. $\endgroup$ Commented Oct 11, 2018 at 20:43


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.