edX: MIT 6.008.1x Computational Probability and Inference Review

MITx: 6.008.1x Computational Probability and Inference is a 12-week probability course offered by MIT on the edX MOOC platform. The first few weeks act as a refresher on fundamental topics in probability, while the bulk of the course focuses on probabilistic graphical models. The last 2 weeks are reserved for a final project. The course lists college calculus and Python as prerequisites; you should be comfortable with mathematical notation because there is a lot of it. The course does not list prior experience with probability as a prerequisite but the general consensus among students is that an introductory course in probability should be a prerequisite. Grading is based on comprehension exercises, weekly homework assignments, 3 programming mini projects and the final project.


Weekly content in 6.008.1x takes the form of a few 5 to 20 minutes lecture videos with accompanying notes that reiterate everything in static written form. The lectures are all written derivations with voice-overs: you never actually see the professor's faces after the first couple orientation videos. The lectures have good information but following along with the professor scribbling out math for 10 minutes or more at a time can get tedious. Chopping long videos into parts and using some clear, typed text to highlight key points in the lectures themselves would be helpful.


Although the course title suggests computation is a central focus, it doesn't come into play much outside of the projects. There are a couple lectures and homework problems that consider data structures and big O complexity, but the computational component is mostly separate from the lectures. The projects require a very solid understanding of the lecture material and sometimes have complicated set ups, which can make it hard to get started and make progress. Mini project difficulty is also very uneven, with the second being much longer and more difficult than the others. The course also makes some questionable design decisions, such as directing students toward using Python dictionaries to store distributions when matrices make the algorithms much easier to implement.


Computational Probability and Inference is a course for students with some prior exposure to probability who want to learn about graphical models and aren't intimidated by walls of mathematical notation. Like other MOOCs by MIT, it is longer than your average MOOC with a good variety of exercises and homework problems, but compared to other MIT MOOCs, the overall course quality falls short. Understated prerequisites combined with rocky programming assignments and a lack of worked example problems are likely to cause a high attrition rate for this course.


I give Computational Probability and Inference 2.5 out of 5 stars: below expectations.


For those with little prior experience with probability (or want to understand it better) and desire the thorough and challenging MOOC experience MIT offers, consider Introduction to Probability - The Science of Uncertainty.