Orange Forum • View topic - (Simple)TreeLearner

(Simple)TreeLearner

A place to ask questions about methods in Orange and how they are used and other general support.

(Simple)TreeLearner

Postby Elaugaufein » Fri Jul 05, 2013 4:11

Hi,

I've been using some Python scripting and attempt to make use of a TreeClassifier and SimpleTreeClassifier and I wanted to ask a couple of questions because I've been getting some results that puzzle me. I've read the documentation and had a look at the py files but since it passes through to C++ its not very useful.

Is the outcome of a TreeLearner / SimpleTreeLearner on a given training set deterministic ie will they always produce the same classifier ?

I ask because the default seeding used for Neural Networks results in Neural Networks being identical unless a different random is passed in than the default. I worked out how to do that with the Neural network but I can't work out how to pass in the appropriate random generator to the (Simple)TreeLearner.

Thanks

Re: (Simple)TreeLearner

Postby Ales » Fri Jul 05, 2013 11:46

Elaugaufein wrote:Is the outcome of a TreeLearner / SimpleTreeLearner on a given training set deterministic ie will they always produce the same classifier ?
The TreeLearner is deterministic, while the SimpleTreeLearner can be randomized by setting the 'skip_prob' attribute to a non zero probability (and set the appropriate 'random_generator')


Return to Questions & Support