hyper-parameters tuning in Lenskit

[manou....@gmail.com]

Dear Michael and Lenskit users,

I am using Lenskit (Lkpy) and I would like to know how you are tuning the hyper-parameters of your recommendation algorithms. Any particular package (i.e., Bayesian optimization)? Is there a published documentation/code that I missed? 

Thank you,

kind regards.

 -- Manel. 

[Michael Ekstrand]

Hi Manel,

I usually use scikit-optimize; it takes a little tweaking to get it working (usually use a custom function to minimize), but I've had good success. I have code to do this in our book-author-gender project: https://github.com/BoiseState/book-author-gender

Adding helper routines to LensKit to make this easier, now that I've debugged it in the context of that project and my class, is on the TODO, probably for LesnKit 0.14. Tracking that work here: https://github.com/lenskit/lkpy/issues/261

- Michael

--
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/2bfbc944-fbbf-4c42-8a0c-97fe59243860n%40googlegroups.com.

--

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.

[Slokom Manel]

Hi Michael,

That's great.

I will have a look at the book author gender project and keep track on issue #261. 

Thank you for the quick reply.

Best,

-- Manel.