I'm searching for a paper or book that list several possible measures for deciding which variable to split a Decision Tree at.
The basic ones are well known and very well documented in several sources:
- Entropy (or Information Gain)
- Gini impurity
- Classification Error
- Variance reduction
But I suppose people come up with all kinds of other split criteria all the time, don't they? Has someone ever collected a list? Preferably with some explanations about the use case and particular advantages (and disadvantages) of the new split criterions.