Mining Massive Datasets

I found the lecture to be of medium difficulty for the post-grad student and I would expect it to be rather hard for an undergrad.

The content is offered in two paces; the lectures of Prof. Ullman are hard to follow, as he browses quickly through many of the notions of the course and does not use enough/ explain in enough detail examples. Jure on the other hand uses a lot of examples and is easy to follow even from an undergrad.

Overall it is a time consuming course, expect to need around 6-8 hours per week. In the end, you do learn quite a few stuff and it is a good lecture to take. I am in favor of the instructors' choice of offering it as it is in Stanford.

Something that could help in the course is to split the content in 10 weeks instead of 7 and add mandatory programming exercises. They help a lot in learning stuck and remembering them for a long time.

Excellent course by the authors, covering the content of the book of the same name http://www.amazon.com/gp/product/1107077230. It is the MOOC version of http://cs246.stanford.edu. Many useful topics in large scale data processing algorithms are covered including mapreduce, pagerank, networks and graph analysis, streaming algorithms, just to mention a few. The level is advanced undergrad or postgrad, with some chapters covering topics in research papers published within the last decade.

Pacing is faster than most other MOOCs (I estimate about 2x the workload of a typical MOOC). But the material is very useful and rewarding. Exercises are comprehensive and the forums are very useful for checking your understanding.