Michael Ekstrand

Also, I don't see clear hooks for normalization right now - should we add those?

Got :+1: from @kluver, will drop this.

We can do this through a deprecation, possibly post-3.0. Keep the metric so old experiments still work, but introduce the new abilities. I would also like to introduce a `rank`...

Thanks for bringing this up! If you want predictions quantized to the nearest valid preference, use `QuantizedRatingPredictor`. For remapping - we do not have a way for item scorers to...

I am personally leaning towards assuming missing scores to be 0.

@jdekstrand and I were pairing on this tonight, and found a difficulty. In the way the scorer is currently defined, we basically have to leave items unscored that cannot be...

> 1) This feels like an implicit problem with ensemble algorithms. Are there standard solutions? The closest thing I know of to a standard solution is to have component produce...

@ell27 I think this would be a good one to work on after #570.

I have started working on this in #872 to give enough to run a script. We need to be able to load models and things like that so that the...

@doxav I had a student who got a long ways w/ LensKit 1.x, there isn't anyone currently working on it. I really want it to happen, and the sooner the...