Funded by the European Union. Views and opinions expressed are however those of the author(s) only and do not necessarily reflect those of the European Union or the European Education and Culture Executive Agency (EACEA). Neither the European Union nor EACEA can be held responsible for them.


Matrix Factorization: A Simple Tutorial and Implementation in Python

Users: 1 - Average Rating: 4.00

This blog post describes in a clear and concise way how the (non-negative and low-rank) matrix factorization technique (Lee and Seung, 1999) can be used in a recommender system to discover the latent features. As a means, author describes an example for recommending films, in which five users rates four films and the unconstrained minimization of the SSE of the representation is accomplished by gradient descent (it could alternatively be computed with an ANN, e.g.). A section on how to regularize the above objective function is included, and a downloadable 23-lines Python implementation with the full source code and this small example along with four interesting references are given as well.


Machine Learning, Optimization

Target Group:

Cite as:
A.A. Yeung. Matrix Factorization: A Simple Tutorial and Implementation in Python. In, Blog Spots section, September 2010.

Author of the review:
Pablo Guerrero-Garcia
University of Malaga


You have to login to leave a comment. If you are not registered click here

Lorenzo Martellini

Nice publication with interesting elements of analysis.