I have enrolled the ML class provided by Stanford last year. But failed due to too much mathmatics, matrix, liner angebra, probabilities and statistics etc. Does this book provide more examples and in a learning by doing style?
I'm working through this book right now. It expects you to know some basic maths - linear algebra, simple probability - if possible, probably at roughly the same level as the Stanford course (I've also been looking at the videos for that), but the book's appendices also provide some quick reviews of these topics if you need them e.g. matrix addition and multiplication. I'm probably at the edge of my own current mathematical comfort zone with these topics, but it's going OK so far.
The main text assumes you know the basics so it doesn't explain these again, although each new ML technique is discussed briefly in theory (with formulae) before being implemented in Python. You can work through the book without understanding all the maths, but it will help you to understand the basics (e.g. matrices and probability as discussed in the appendices), so you can understand what the Python code is actually doing. I've actually found it harder (so far) figuring out what the calls to Numpy and matplotlib functions are doing - you may want to check the documentation for these packages as the book doesn't explain much of this stuff.