Learn R Programming

poismf (version 0.4.0-4)

Factorization of Sparse Counts Matrices Through Poisson Likelihood

Description

Creates a non-negative low-rank approximate factorization of a sparse counts matrix by maximizing Poisson likelihood with L1/L2 regularization (e.g. for implicit-feedback recommender systems or bag-of-words-based topic modeling) (Cortes, (2018) ), which usually leads to very sparse user and item factors (over 90% zero-valued). Similar to hierarchical Poisson factorization (HPF), but follows an optimization-based approach with regularization instead of a hierarchical prior, and is fit through gradient-based methods instead of variational inference.

Copy Link

Version

Install

install.packages('poismf')

Monthly Downloads

225

Version

0.4.0-4

License

BSD_2_clause + file LICENSE

Issues

Pull Requests

Stars

Forks

Maintainer

David Cortes

Last Published

March 26th, 2023

Functions in poismf (0.4.0-4)

topN.new

Rank top-N highest-predicted items for a new user
factors

Determine latent factors for new rows/users
topN

Rank top-N highest-predicted items for an existing user
poismf_unsafe

Poisson factorization with no input casting
get.factor.matrices

Extract Latent Factor Matrices
predict.poismf

Predict expected count for new row(user) and column(item) combinations
print.poismf

Get information about poismf object
get.model.mappings

Extract user/row and item/column mappings from Poisson model.
poismf

Factorization of Sparse Counts Matrices through Poisson Likelihood
factors.single

Get latent factors for a new user given her item counts
summary.poismf

Get information about poismf object