This course, which is designed to serve as the first course in the Recommender Systems specialization, introduces the concept of recommender systems, reviews several examples in detail, and leads you through non-personalized recommendation using summary statistics and product associations, basic stereotype-based or demographic recommendations, and content-based filtering recommendations.
After completing this course, you will be able to compute a variety of recommendations from datasets using basic spreadsheet tools, and if you complete the honors track you will also have programmed these recommendations using the open source LensKit recommender toolkit.
In addition to detailed lectures and interactive exercises, this course features interviews with several leaders in research and practice on advanced topics and current directions in recommender systems.
Week 1: Preface
This brief module introduces the topic of recommender systems (including placing the technology in historical context) and provides an overview of the structure and coverage of the course and specialization.
Week 2: Non-Personalized and Stereotype-Based Recommenders
In this module, you will learn several techniques for non- and lightly-personalized recommendations, including how to use meaningful summary statistics, how to compute product association recommendations, and how to explore using demographics as a means for light personalization. There is both an assignment (trying out these techniques in a spreadsheet) and a quiz to test your comprehension.
Week 3: Content-Based Filtering -- Part I
The next topic in this course is content-based filtering, a technique for personalization based on building a profile of personal interests. Divided over two weeks, you will learn and practice the basic techniques for content-based filtering and then explore a variety of advanced interfaces and content-based computational techniques being used in recommender systems.
Week 4: Content-Based Filtering -- Part II
The assessments for content-based filtering include an assignment where you compute three types of profile and prediction using a spreadsheet and a quiz on the topics covered. The assignment is in three parts -- a written assignment, a video intro, and a "quiz" where you provide answers from your work to be automatically graded.
In this course, you will learn the fundamental techniques for making personalized recommendations through nearest-neighbor techniques. First you will learn user-user collaborative filtering, an algorithm that identifies other people with similar tastes to a target user and combines their ratings to make recommendations for that user. You will explore and implement variations of the user-user algorithm, and will explore the benefits and drawbacks of the general approach. Then you will learn the widely-practiced item-item collaborative filtering algorithm, which identifies global product associations from user ratings, but uses these product associations to provide personalized recommendations based on a user's own product ratings.
Week 1: Preface
Note that this course is structured into two-week chunks. The first chunk focuses on User-User Collaborative Filtering; the second chunk on Item-Item Collaborative Filtering. Each chunk has most of the lectures in the first week, and assignments/quizzes and advanced topics in the second week. We encourage learners to treat each two-week chunk as one unit, starting the assignments as soon as they feel they have learned enough to get going.
Week 2: User-User Collaborative Filtering Recommenders Part 2
Week 3: Item-Item Collaborative Filtering Recommenders Part 1
Week 4: Item-Item Collaborative Filtering Recommenders Part 2
In this course you will learn how to evaluate recommender systems. You will gain familiarity with several families of metrics, including ones to measure prediction accuracy, rank accuracy, decision-support, and other factors such as diversity, product coverage, and serendipity. You will learn how different metrics relate to different user goals and business goals. You will also learn how to rigorously conduct offline evaluations (i.e., how to prepare and sample data, and how to aggregate results). And you will learn about online (experimental) evaluation. At the completion of this course you will have the tools you need to compare different recommender system alternatives for a wide variety of uses.
Week 1: Preface
Week 2: Advanced Metrics and Offline Evaluation
Week 3: Online Evaluation
Week 4: Evaluation Design
In this course you will learn a variety of matrix factorization and hybrid machine learning techniques for recommender systems. Starting with basic matrix factorization, you will understand both the intuition and the practical details of building recommender systems based on reducing the dimensionality of the user-product preference space. Then you will learn about techniques that combine the strengths of different algorithms into powerful hybrid recommenders.
Week 1: Preface
Week 2: Matrix Factorization (Part 1)
This is a two-part, two-week module on matrix factorization recommender techniques. It includes an assignment and quiz (both due in the second week), and an honors assignment (also due in the second week). Please pace yourself carefully -- it will be difficult to finish in two weeks unless you start the assignments during the first week.
Week 3: Matrix Factorization (Part 2)
Week 4: Hybrid Recommenders
This is a three-part, two-week module on hybrid and machine learning recommendaton algorithms and advanced recommender techniques. It includes a quiz (due in the second week), and an honors assignment (also due in the second week). Please pace yourself carefully -- it will be difficult to finish the honors track in two weeks unless you start the assignments during the first week.
This capstone project course for the Recommender Systems Specialization brings together everything you've learned about recommender systems algorithms and evaluation into a comprehensive recommender analysis and design project. You will be given a case study to complete where you have to select and justify the design of a recommender system through analysis of recommender goals and algorithm performance.
Learners in the honors track will focus on experimental evaluation of the algorithms against medium sized datasets. The standard track will include a mix of provided results and spreadsheet exploration.
Both groups will produce a capstone report documenting the analysis, the selected solution, and the justification for that solution.
