Recommender System With Machine Learning and Statistics

Recommender system is a promising approach to boost sales to the next level by suggesting the right products to the right customers.

This course starts by showing you the main solutions of recommender systems in the industry and the hypotheses behind the main solutions. You’ll then learn how to build collaborative filtering models with fast.ai, and exercise the trained model on test datasets.

As you advance, you’ll visualize latent features, interpret weights and biases, and check what similar users/Items are from the model’s perspective. Furthermore, you’ll build a hybrid recommender system with popularity and association rule, and evaluate the recommendations with selected criteria.

By the end of this course, you’ll be able to explain the theories and assumptions of recommender systems and build your own recommender on other datasets using python.

The outline of course is as follows:

• Why Business Needs Recommender Systems

• Roadmap of the Course

• The Hypotheses Behind the Main Solutions of Recommender Systems

• Hands-on Collaborative Filtering Recommender System With Fast.ai on Instacart Grocery Dataset

  • A Quick Eda on the Grocery Dataset

  • What Is Collaborative Filtering in Depth

  • How to Build and Train Collaborative Filtering Model With Fast.ai

  • How to Visualize Latent Features? Do Popular Items Have a Higher Bias? What Are Similar Users From Model Perspective?

• Step-By-Step Guide to Build a Hybrid Recommender System With Popularity and Association Rule

  • What Is the Definition of Popularity and What Is Support

  • How to Encode an Item-Order Matrix

  • What Are Confidence and Lift

  • What Is Association Rule and How to Apply Apriori Algorithm

  • How to Evaluate Results With Selected Criteria

• End-Of-Course Conclusion