MultiEval support

[Michael Ekstrand]

Hi LensKitters,

Quick question - is anyone actually using the MultiEval class in LensKit for Python? What do you like about it?

I initially created it as a hopefully-easy way to do parameter sweeps and things, but in practice never actually use it myself - I just write Python loops to train and evaluate my models on a data set. All the quick-start docs recommend this now as well.

Is there a compelling reason to keep it? "Explicit is better than implicit", and directly writing Python loops seems easy enough. Keeping it around induces maintenance and testing overhead.

If it isn't useful, I'd like to drop it before we cut our non-0.x release. I don't have a firm timeline on that, but feel like we've gotten enough experience with the software that later this year seems feasible. We'll jump from 0.x to 4.0 - this replaces LensKit for Java, which was discontinued when we were working on 3.0.

- Michael

--

Michael D. Ekstrand — michael...@boisestate.edu — https://md.ekstrandom.net
Assistant Professor, Dept. of Computer Science, Boise State University

People and Information Research Team (PIReT) — http://coen.boisestate.edu/piret/

I may send mail outside of working hours; I do not expect you to. He/him.

[Daniel Kluver]

I can say with confidence that I havn't used it. Without looking deep into the design alternatives it's hard to be sure which is truly better, but I expect that if you can't think of something it does better than a for loop, then we probably don't need it.

~Kluver

--
You received this message because you are subscribed to the Google Groups "LensKit Recommender Toolkit Development and Support" group.
To unsubscribe from this group and stop receiving emails from it, send an email to lenskit-recsy...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/lenskit-recsys/CABSXaUvfE6NLAFome0NaWsvRhcDzQ46J9yqqbvdRLLkF4rcHjQ%40mail.gmail.com.