Re: [Scikit-learn-general] [scikit-learn] Weighted pls -- adding support for sample_weight option. (#4974)

[Deepak Subburam]

Hello, can someone (Gael?) familiar with PLS regression, or SVD, review my pull request #4974 which adds support for sample_weight option to the PLS regression classes? Andreas has reviewed the style and test issues with my change, which I have since addressed, but is not comfortable reviewing the math parts of the change (message below). The math parts would be at line 86 and lines 34-35 of pls_.py. Direct link to the pull request: https://github.com/scikit-learn/scikit-learn/pull/4974 Some context: I and my cofounder run a small data science consulting company, and used this module for a client project. We find results more robust once we add sample_weights, and it'd be great if the client can use the master scikit-learn branch to run the code rather than our branch. Thank you, Deepak On 7/30/2015 1:07 PM, Andreas Mueller wrote: Thanks for the contribution. Unfortunately I'm not really familiar enough with PLS to review the changes. It seems after the refactoring the code got quite a bit longer. Is that mostly due to docs? There are a bunch of test failures because some classes are now missing a predict method. — Reply to this email directly or view it on GitHub. Date: Tue, 7 Jul 2015 02:03:41 +0200 From: Gael Varoquaux<gael.varoqu...@normalesup.org> Subject: Re: [Scikit-learn-general] Planning on implementing     sample_weight option for PLSRegression.fit() To:scikit-learn-general@lists.sourceforge.net Message-ID:<20150707000341.ga2996...@phare.normalesup.org> Content-Type: text/plain; charset=us-ascii On Mon, Jul 06, 2015 at 04:55:07PM -0400, Deepak Subburam wrote: I have some time to contribute, and would like to implement a sample_weight option to the fit() method of sklearn.cross_decomposition.PLSRegression, handling any other knock-on effects. Let me know if you think this is not a good idea, or have any other thoughts to guide me. In general, I think that this is a great idea. One question is: is the quality of the PLS code ready for these changes? G

------------------------------------------------------------------------------

_______________________________________________
Scikit-learn-general mailing list
Scikit-learn-general@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/scikit-learn-general