Practical Machine Learning on H2O

I started off with Andrew Ng's classic Machine Learning course and was happy to check out other Coursera offerings, but this one was definitely a disappointment.

This course starts off with the prerequisite of having to know either R or python, and some deceptively basic statistical concepts like distribution, bias/variance, confusing matrix, etc. Working from this, it's incredible that the instructor, Darren, rams through a few models of unsupervised learning in week 5 of 6. When it actually comes to explaining how the underlying algorithms work, it's poorly explained, and the student is left with click on this wikipedia page, etc.

Other 1-star activity:
Darren routinely flips between R and python h2o, in the videos- confusing and not helpful.

Darren has never posted to the discussion forum, so asking a basic question like where is the code repository that he offered, goes unanswered (you're left typing the code by pausing/forwarding his videos).

Darren did not complete the second question on the final assignment. Only one of the two questions are complete; the classification problem is unfinished. Amazingly, for the first question, he provides the data, without actually explaining what all of the columns mean (wow!). After completing the assignment, it appears that the data might have come from a prior kaggle competition.

The coursera platform is flawed, as students are left to marking each other's assignments. This means if not enough people are enrolled, you will have to re-enroll for the next stream, in hope that you get all of your assignments marked.