By: Aaron Berk
This is a whirlwind introduction to matrix completion theory and methods. It was originally created for the 2017 BC Data Science Workshop. Some modifications have been made to protect proprietary data and heretofore unpublished methods.
This tutorial includes several exercises. Some of these exercises are novel — you coudl be the first to answer them!
The notebook in this tutorial requires the cvxpy package, and relies
on the splitting cone solver SCS, which is part of
cvxgrp. For installation details,
visit the cvxpy
website, which should cover the automatic installation of SCS and other solvers.
This code also requires the latest versions of sklearn and scipy.
I have attempted to ensure that all appropriate references are given in-line in the notebook. If it seems I have made an omission, please e-mail me or submit a pull request.
I would like to thank Oscar Lopez for his help and comments with some of the finer points of this tutorial.