MatrixFactorization
A series of tutorials on the uses of matrix factorization in data science. The tutorial is in four parts:
- Part 0: The Singular Value Decomposition: This notebook introduces the singular value decomposition, and shows its usefulness in dimensionality reduction and image compression.
- Part 1: The Jester Dataset: This notebook introduces the Jester dataset, which contains user ratings for a set of jokes. This dataset will be used in parts 2 and 3 to build a recommendation system. This notebook also allows the user to rate a subset of the jokes in the dataset to allow the recommendation systems developed in later parts to recommend jokes based on the user's taste.
- Part 2: Alternating Least Squares: This notebook introduces the Alternating Least Squares algorithm, and uses this algorithm to build a recommendation system for the Jester dataset introduced in part 1.
- Part 3: Gradient Descent for Matrix Factorization: This notebook introduces the gradient descent method for building a recommendation system, and uses this algorithm to build a recommendation system for the Jester dataset introduced in part 1.
The notebooks are intended to be run in sequential order.