I recently modified the kernel density estimation routines in sklearn/neighbors to include optional weighting of the training samples (to make analogs to weighted histograms). I'd be interested in contributing this to scikit-learn, but it's mostly edits of existing code (as opposed to new source files), and I'm not sure what's the policy in that case.
Here's the code, which could use some more testing and validation (and documentation): https://github.com/jaredgabor/scikit-learn/tree/weighted_kde Thanks for any input, Jared Gabor
------------------------------------------------------------------------------
_______________________________________________ Scikit-learn-general mailing list Scikit-learn-general@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/scikit-learn-general
