I'm just starting to get back to data science and glad to have found Hadley Wickham's and Garrett Grolemund's book, R for Data Science. It looks to be a good refresher for me.

In reading the Introduction, I'm immediately struck with how helpful this book is going to be. In addition to refreshing my knowledge, it is written with a large degree of instructional intent.

Section 1.3, What you won't learn, is a very specific and thoughtful description of the topics that are not covered. The one that stood out for me has to do with Hypothesis Confirmation. The authors point out that confirmation of a hypothesis requires precise model and the creation of an analysis plan that assures that you do not look at each observation more than once. Multiple views of an observation means you are doing exploratory data analysis, which is part of hypothesis generation.