He makes it look easy!
It is an 11 week course taught by Dr. Andrew Ng, Stanford University(co-founder of Coursera). The course starts from absolute scratch and hence only a very basic mathematical background is all that is required in terms of prerequisites. Almost every week has a programming assignment which is graded out of 100 along with a MCQ style test testing some theoretical as well as practical knowledge of the student.
The course covers Supervised Learning, Unsupervised Learning, SVM, Neural Networks, Anomaly Detection, Recommender Systems, Online Learning and many other facets of Machine learning. Apart from this, Prof Andrew Ng provides in-depth knowledge of the approach that should be followed in terms of implementing a machine learning solution on a data set. Throughout the course, one would often hear that he always prefers a quick and dirty implemetation of the problem at hand, and then plot helpful learning curves to guide his decision further in improving efficiency and accuracy. In terms of saving time , the advise indeed is invaluable as often times, building a large training set for months is not the best way to improve accuracy on the Test Set.
The only lesson where he struggles to get his thoughts across is the back propogation chapter, but he gives enough time for the topic to sink in by the end of the course.
In terms of completing the course, i feel like 4-6 weeks is quite enough time, if done regularly. Off to CS231n :).