predict_proba method and n_more_iter functionality for MCMC Classification about fastfm

[ibayer]

For MCMC Classification there is only one fit_predict method which is not practical for an productional use (because its necessary to train the model each time one want to predict an example).

It is always the case for MCMC (applied to non-linear models) that the predictions can only be calculated while training the model.

Have a look at our paper Sample selection for MCMC-based recommender systems for more details, you can get it here without the pay-wall.

Also the warm_start functionality of the other solvers is really nice but not available for this solver.

True, it's only available for mcmc regression. It should be possible to implement the classification using the same approach. Are you still interested after learning that mcmc models are hard to use in production? ;-)