Lady Luck: The Theory of Probability
By Warren Weaver.
I’m on a statistics kick. I want to learn the fundamentals such that I can onto machine learning with a solid base of from whence it came. I wasn’t sure exactly how I wanted to do this, so I’ve been doing a combination of reading books, flipping through textbooks, taking online courses (most of which I don’t like and don’t finish, but that’s a different story,) and researching explanations on Wikipedia. This is the first book I finished as part of this endeavor.
I like it’s style, but I found the content to be a little lateral to what I was looking for. It’s from the sixties and also (obviously) focused more on probability than statistics, though the two are strongly related.
I enjoyed the mix of narrative and technical explanation. Weaver doesn’t shy away from equations and uses standard notation for referring back to previous ones. He wrote it for precocious high schoolers and I think fairly accurately judged that level: as someone with an undergraduate education I found the math straightforward. Which is not to say it was unnecessary; I was extremely glad he got detailed with the math and was explicit about when he was not explaining a proof because it was too complicated or detailed or, and generally this was in combination with the previous, not particularly necessary.
The narrative was useful in tying the theory to its history and uses. However this is where the book was a little astray for me. Weaver talks mostly about counting probabilities, a lot of which came from gambling games. While I found it interesting, it definitely was not the kind of statistics I was looking for. He didn’t connect it to experimental design and hypothesis testing at all. It’s really a different side of statistics that I didn’t realize wasn’t, at least in this book, entirely related.
Still, I enjoyed the book and its contents and especially its style.