I believe it would fit in sklearn-contrib even if it's more for statistical inference rather than machine learning style prediction.

Others might disagree.

Anyways, joining efforts to improve documentation, CI, testing and so on is always a good thing for your future users.

--

Olivier