Hey folks, here's my request for project inclusion in scikit-learn-contrib.

• Project name: polylearn
• Project description: factorization machines and polynomial networks in Python (implements the solvers from @mblondel et al's ICML 2016 paper.
• Authors: Vlad Niculae (@vene)
• Current repository: https://github.com/vene/polylearn
• Requirements:
  • scikit-learn compatible (check_estimator passed) (see notes)
  • Documentation (guide, API reference, example gallery) (see notes)
  • Unit tests (coverage: 98%)
  • Python3 compatible
  • PEP8 compliant (I should really add CI for this...)
  • Continuous integration

Notes:

• There could certainly be more narrative docs and more examples. I will work on that.
• The two regressor classes pass check_estimator, but the classifier classes do not. The only reason for this is because polylearn only does binary classification, raising an error on multiclass or multioutput targets. The common tests don't like this, which I think is too strict. Should we make the scikit-learn tests more leninent with this scenario?

Cheers,
Vlad