Exercises and lessons from the book Essential Math for Data Science by Thomas Nield. Mostly just creating this repo to force myself to work through this book.
Chapter 1 This chapter does a good job setting the tone for the book. If you are comfortable with the level of detail here, buckle up, this is the book for you! If you want more, you should keep going for a few more chapters. The first chapter of stats books is always a little rough, but this is one of the better ones as far as introducing the reader to the broad sweep of math and stats, as well as python and data science broadly.
Chapter 2 The probability chapter. I like that this comes right after the introductory chapter since it's key to data science. Kudos to the author for explaining multiple courses in stats in just 20 pages or so. The descriptions of distributions and how you code those out, as well as use them for probabilistic thinking, is maybe the best I've read. I probably would have gotten better grades in grad school has I read this a decade ago.
Chapter 3 The formal intro to statistics. This chapter is kind of fun and builds on the previous chapter on probability.
Chapter 4 Vectors are tough to explain without showing the reader what you're talking about. This chapter attempts that, but felt like a bad Medium post at times in that it occasionally falls into using jargon to explain topics rather than getting at their underlying point. Overall, an important, often glossed over topic in data science that I'm glad to get a refersher on.
Chapter 5 I read this chapter and then went back and quickly ran through some of the examples and exercises many weeks later. On one level, this chapter, like chapter 1 of every stats book where you learn about mean and mode, felt slightly perfunctory, but it's a brief chapter en route to the juicy stuff. Also, it was fun (in the most messed up way possible for a data science book) to revist the formulas that regression is built on instead of assuming the folks at scikit-learn and every other package got it right. On a basic level, it gives you an appreciation for just how much goes into the basic functions of the major packages. This chapter is great, though, for showing how each of the previous chapters build up to actually fitting a model.
Chapter 6 A fun chapter that provides a nice refresher on logistic regression. This was probably as good as you can get with a "gentle introduction" without going down any number of forking paths related to logistic regression. The best part of this chapter is focusing on the statistics underlying logistic regression. Similar to Chapter 5, it's easy to forget what you are actually doing when you run all of the pre-packaged functions from different libraries.
Chapter 7 This is a nice intro to neural nets in that it shows how complicated they can get very quickly. Building a neural net from scratch, like fitting a logistic regression from scratch, is overkill in any context except one where you want to really learn. I wouldn't have worked through that without this book, so it's a nice thing I hope I don't have to do again. Overall, a good note to end on. The exercise dataset is also a good example of how neural nets don't necessarily outperform other models.