What will be the parameter range of decision tree in optimized operator.

vbsingh

What will be the parameter range of decision tree in optimized operator and which parameter should be chosen for optimized result. There are so many parameters in decision tree, namely criterion, minimal size for split, minimal leaf size, minimal gain, maximal depth, confidence and number of prepruning alternatives.

varunm1

Hello @vbsingh

Parameters ranges are available in the help window or decision tree in rapidminer. If you want to see how others chose (Wisdom of Crowds) that you can click on the green arrow mark as shown in the below image.


You can see from this image most of the users chose values between 20 and 29 and some chose between 10 and 19, so you can set the range between 10 and 30. Similarly for others as well. If you are unable to see this kind of recommendation, it means that you did not activate the wisdom of crowds, you can see that at the bottom of the rapidminer window. If you are unable to find it there, you can also go to SETTINGS --> PREFERENCES --> Recommender and then check the "enable operator recommendations".

You should also look at the definitions of the parameters to see how they impact a decision tree for better hyperparameter tuning. One important thing is that if you use criterion in optimize parameters for classification problem be sure to remove the "least squares" criterion from parameter search as it is related to regression.

Hope this helps.