recommender-als: Recommendations using alternating least squares algorithm

[ bsd3, library, numeric ] [ Propose Tags ]

This package provides a recommendation algorithm based on alternating least squares algorithm, as made famous by the Netflix Prize.

It takes as its input a list of user-item pairs and returns a list of recommendations for each user. The current implementation is limited to using unrated pairs.

The algorithm is parallelized and should be quick enough to train the model within seconds for a few thousand users and items. Getting recommendations from a computed model happens nearly instantly.

For implementation details, see "Large-scale Parallel Collaborative Filtering for the Netflix Prize" by Yunhong Zhou, Dennis Wilkinson, Robert Schreiber and Rong Pan.

Versions [faq] 0.1.0.0, 0.2.0.0, 0.2.1.0, 0.2.1.1
Change log ChangeLog.md
Dependencies base (>=4.11 && <5), containers (>=0.5 && <1), data-default-class (>=0.1.2 && <1), hmatrix (>=0.20 && <1), parallel (>=3.2 && <4), random (>=1.1 && <2), vector (>=0.11 && <1) [details]
License BSD-3-Clause
Copyright Kari Pahula 2020
Author Kari Pahula
Maintainer kaol@iki.fi
Category Numeric
Home page https://gitlab.com/kaol/recommender-als
Uploaded by kaol at 2020-07-09T07:43:48Z
Distributions NixOS:0.2.1.1
Downloads 206 total (124 in the last 30 days)
Rating (no votes yet) [estimated by Bayesian average]
Your Rating
  • λ
  • λ
  • λ
Status Hackage Matrix CI
Docs available [build log]
Last success reported on 2020-07-09 [all 1 reports]

Modules

[Index] [Quick Jump]

Downloads

Maintainer's Corner

For package maintainers and hackage trustees