Writing tests to gain faimliarity

[Saikat Kanjilal <sx...@hotmail.com>]

Hi Mahout committers,I'd like to gain a lot more familiarity with some of the clustering algorithms in mahout as well as naive bayes and random forest and was thinking that one good way might be to add more tests or refactor some of the currently existing tests, in looking through JIRA I didnt see anything related to this.   Is there value added in doing this based on currently existing tests, if not it'd be great to get suggestions in 1 or more areas where I can contribute tests as a newish person.  Also in addition to tests another area I'd like to help out on is in the implementation of a support vector machine, I was wondering if anyone is planning or is going to submit their own SVM implementation into mahout?
Regards 		 	   		  

Re: Writing tests to gain faimliarity

[Robin Anil <ro...@gmail.com>]

Tests+++. Creating toy dataset expectations across multiple classifier will
be a great thing. Right now we don't track if there are regressions in the
codebase with integration tests. You could maybe try and create a set of
smallish datasets and run each classifier, spit some metrics and put
expectations around them.