what is the usage of parameter “confidence” in operator “Decision Tree”？

lg21c

what is the usage of parameter “confidence” in operator “Decision Tree”？

yyhuang

Hi @lg21c,

Not sure about the definition of f and q in your probability function. But you can refer to the source code of decision tree and pessimisticpruner.java scripts here
https://github.com/rapidminer/rapidminer-studio-modular/blob/master/rapidminer-studio-core/src/main/java/com/rapidminer/operator/learner/tree/DecisionTreeLearner.java

https://github.com/rapidminer/rapidminer-studio-modular/blob/master/rapidminer-studio-core/src/main/java/com/rapidminer/operator/learner/tree/PessimisticPruner.java

If the tree grows too big, the model is easily overfitting. The confidence level is used to determine whether or not prune the branches based on pessimistic errors.



HTH!

YY

lg21c

Hi @yyhuang , it would be a lot easier for people to know the complex parameters if rapidminer company provide detailed technical manual，many users can't read java

lg21c

The formula in the program is complicated，I just want to know

yyhuang

This operator for decision tree is the same as Quinlan's C4.5 or CART depending on the criterion, e.g. using gain_ratio or Gini.
Useful reference
Quinlan,J.R.: C4.5: Programs for Machine Learning Morgan Kauffman, 1993