@erictleung a wow! book... excellent exposure of some elemental probability distributions from a more geometrical point of view and sampling theory. Unusual: normally the emphasis is purely analytical. This approach is more from physics. Are the authors physicists? This approach is also very useful to understand multidimensionality, which is the key of many Data Science problems and understanding random point generators.
All the first part of the book emphasises parametric statistics, particularly Gaussian.
The same approach (Gaussian, Vector Space, etc) goes for the second chapter but this is not unusual. SVD is a VERY important transformation, and it is key for many other ones in parametric statistics.
Then, for the graph part, some welcome combinatronics. Good! The random generation of triangles in a graph and then the derivation of the variance of a graph based on them is wow for its simplicity! Excellent. And then transitions... those guys are physicists... This is the chapter I am more interested...
Analysing Markov theory after graphs is a BIG plus. Definitively a approach based on space representation. @erictleung: relevant for you if you want to go Biotech this chapter.
And it is not until all that theory that you start the Machine Learning part... OK!!! Good! Ending with algorithms for big data...
@erictleung this book is a PIECE OF WORK. If you have ever the money for the finished book, BUY IT! Pity doesn't consider some programming exercises, but it is a well written book.