A Review of Machine Learning Foundations — A Case Study Approach by Coursera

On December 11, 2016 I completed the course “Machine Learning Foundations: A Case Study Approach” by Coursera. This course is a great introduction to the world of Machine Learning, and through this blog post my goal is to give a brief review of the course and its content.

What is it About?

“Machine Learning Foundations: A Case Study Approach” is an introductory course about common Machine Learning concepts such as regression, classification, clustering and similarity, recommender systems, and deep learning. It’s a hands-on-experience course (what they call a use-case study) which allows for a more practical understanding of common methods used in Machine Learning, rather than diving up immediately to the theory behind them. The course serves as a foundation to getting started with the Machine Learning specialization, which will later cover those same topics in more detail.

Structure and Content

The course is structured in 6 weeks (about 10 hr per week commitment), each of them covering a specific Machine Learning concept. Each Machine Learning concept is explained through a series of video lessons followed by a quiz (usually 5–10 questions), and finally a programming assignment in which you will implement a small application using the Machine Learning method studied during that course week. Lessons use an approach which is more focused on general principles rather than specific implementations or tools of it.

Instructors recommend using Python with GraphLab (a Machine Learning modeling tool for developers and data scientists), but other languages or packages can be used as well. There’s no need to install the recommended packages on your local machine, since a GraphLab service running on Amazon’s cloud is already provided for each student enrolled in the course.

Conclusion

I enjoyed taking this course, it’s an introductory course and thus you might be able to skip it if you already have some experience with Machine Learning. Immediately after completion, I started taking the Regression course, and as explained above, this course is a lot more theoretical and algorithms will now be implemented from scratch (instead of using third party libraries).