Evaluation of Classification and Regression Tree

Hey guys,

I am new to Knime recently for my academic degree.

Right now, i am learning building Classification and Regression Tree models in Knime.

But how to evaluate those models? Except confusion matrix, any other method?

 

Many thanks,

 

HW

Hi,

Lots of methods, depends on what exactly you're after in the evaluation. Assuming it's mainly predictive accuracy/adequacy, I'd suggest 10 times 10-fold cross-validation.

For other options/more background there's a good free ebook here:

http://www-stat.stanford.edu/~tibs/ElemStatLearn/

Additionally I'd recommend the following two classics, the second one strategically oriented, and remarkably open to CART as well:

http://www.amazon.com/Data-Mining-Practical-Techniques-Management/dp/0123748569/ref=sr_1_1?ie=UTF8&qid=1379592602&sr=8-1&keywords=data+mining

http://www.amazon.com/Regression-Modeling-Strategies-Applications-Statistics/dp/0387952322/ref=sr_1_1?ie=UTF8&qid=1379592760&sr=8-1&keywords=regression+strategy

Cheers
E

Thanks very much for your help!

Very useful!

In adiition, I am not sure how to do "pruning with validation data" in Knime when building regression tree , do you have any idea on this?

 

Appreciated

HW