search

alex-pirozhenko / sklearn-pmml

A library that allows serialization of SciKit-Learn estimators into PMML
MIT License
70 stars 17 forks source link

JPMML Integration Tests + Random Forest Classifier #9

Closed NeverNude closed 8 years ago

NeverNude commented 8 years ago

Hey, made a couple of changes.

First is that I included a small Java project and added some JPMML integration tests. The JPMML jar linked in the POM is from an earlier version of JPMML that is BSD licensed so that the AGPL won't affect people that want to update this on their own. I added integration tests for Decision Tree Classifier/Regressor and the RandomForestClassifier. It should be straightforward to add for the GradientBoostingClassifier. Right now all the integration test does is generate 10 numerical features and compares the output from the sklearn model to the jpmml evaluator. I will extend it later to support categorical features. The tests will log a warning and pass if someone doesn't have java and or maven installed.

alex-pirozhenko commented 8 years ago

Please merge from the latest master - I've added java and maven to travis.yml, so the JPMML test should work now

NeverNude commented 8 years ago

OK great, will do

On Tue, Aug 18, 2015 at 8:13 AM, Alex Pirozhenko notifications@github.com wrote:

Please merge from the latest master - I've added java and maven to travis.yml, so the JPMML test should work now

— Reply to this email directly or view it on GitHub https://github.com/alex-pirozhenko/sklearn-pmml/pull/9#issuecomment-132245341 .