It works for me. And my computer's fine =P Actually that's the only mirror that still open. Of course it's illegal. Didn't have to post it. But there are other link if interested (.org)
It will be pretty tough to follow much of the ML literature without at least a basic understanding/exposure to vectors and matrices. Ng. does a great job of providing a preliminary understanding of how to read and apply much of the required notation. One of the great features of this course is that it allows you to get a much better concrete and intuitive grasp of the concepts via excellent programming examples. I thought this course could be taken any time - as they moved to open enrollment some time ago.
This is a solid class. Machine learning was the one class I had in grad school that was badly taught. This version of the class is much more solidly taught. However it's worth mentioning that the homework isn't really that long or difficult, and the amount of implementation you do is fairly minimal.
This isn't a "how to trade" course, it's not meant to be a tutorial. If you want to understand a subject like this you have to learn some base theory with no immediate practical application. That's kind of the basis for almost every advanced field in the world, you don't just skip the anatomy class to start operating on live patients because you weren't getting anything practical out of it. This is a great course if you're interested in truly understanding the subject taught by one of the best in the field at no cost. What's not to like about that!
It's a good introduction to machine learning but I don't see what the course has to do with trading. Maybe to teach you to look elsewhere for automated trading? Just sayin' ... (yes, I took the course, several years ago and it's a good intro but nothing more)
I was discouraged by the use of Octave in that Andrew Ng class. I am in the last of 10 weeks in Statistical Learning from Stanford also. They use R which is much preferred. Tough course. They used S&P500 index prices to demonstrate how a predictive model using logistic regression works which is quite cool. Overall though these models appear be most useful very very big data like bioinformatics and gene data modelling where you can have thousands of X's (predictive variables). I've already got a passing grade but not a convincing one. High quality course and free text. Here's the link https://lagunita.stanford.edu/courses/HumanitiesSciences/StatLearning/Winter2016/