Hands-On Recommendation Systems with Python

Singular value decomposition

Singular value decomposition, or SVD, is a type of matrix analysis technique that allows us to represent a high-dimensional matrix in a lower dimension. SVD achieves this by identifying and removing the less important parts of the matrix and producing an approximation in the desired number of dimensions.

The SVD approach to collaborative filtering was first proposed by Simon Funk and proved to be extremely popular and effective during the Netflix prize competition. Unfortunately, understanding SVD requires a grasp of linear algebraic topics that are beyond the scope of this book. However, we will use a black box implementation of the SVD collaborative filter as provided by the surprise package in the next chapter. ...