Project: Recommender Engines

Introduction

Many machine learning algorithms, including some used in recommender engines, require two phases in order to be effective.

1. A Training Phase: All data about all customer’s preferences are combined, and the algorithm creates a model of how it views the world. This may include groupings of what customers are similar to each other, or just internal data structures it can use to quickly generate a prediction later.
2. A Query Phase: The training phase often requires a lot of time and memory, whereas the query phase is generally pretty fast.

General Techniques

There are three techniques used generally in recommender engines

• Market Basket Analysis (MBA)
• Collaborative Filtering (CF) — ML, supervised learning
  • user based
  • item based
  • content based
• K-Means Clustering (KMC) — ML, unsupervised learning

Programming Experience

Let us use Python for this project!

References

• Paper: An Algorithm Framework for Performing Collaborative Filtering

Statistics

###

ProTip:** TBU