Statistics by Freedman, Pisani and Purves. Don't know if I got better but loved the real world examples and cartoons. Does not have too many pre-requisites. Each section presents a tiny concept which is followed by plenty of exercises that have answers at the end. The furthest I got in a book in recent days, Math or not.
I taught from this book (it wasn't my choice, it was the standard book where I was teaching). It's really good for intuition, but because it doesn't use standard notation I think it might have done a disservice to students who were going to go on to learn more.
My pandemic project in 2020 was to finally read through the used copy I bought a decade ago. I agree it was really useful at building foundational intuitions. And that it doesn't use professional jargon which sometimes makes Stats Wikipedia's "death by integrals" approach a dense barrier to entry.
For example, the book uses "the box model" all over the book but is not used anywhere else, and every else uses the phrase "i.i.d" which is not used in the book.
Still, it's been really useful at my job in reasoning about timeseries data from Prometheus, especially in canary analysis. Far more useful than the whirlwind tour of distributions my 1 semester "Statistics for Engineers" course in college undertook.
Any suggestions for more conventional Statistics books? (Math-oriented, with proofs if possible). I'm reading Devore's one for Engineering and the Sciences and it's pretty good but I'm having a bit of hard time with p-values, hypothesis tests, etc. and wanted a second book as reference for those topics.
