Here are my first impressions of The Art of R Programming. I haven’t had time to read it thoroughly, and I doubt I will any time soon. Rather than sitting on it, I wanted to get something out quickly. I may say more about the book later.
The book’s author, Norman Matloff, began his career as a statistics professor and later moved into computer science. That may explain why his book seems to be more programmer-friendly than other books I’ve seen on R.
My impression is that few people actually sit down and learn R the way they’d learn, say, Java. Most learn R in the context of learning statistics. Here’s a statistical chore, and here’s a snippet of R to carry it out. Books on R tend to follow that pattern, organized more by statistical task than by language feature. That serves statisticians well, but it’s daunting to outsiders.
Matloff’s book is organized more like a typical programming book and may be more accessible to a programmer needing to learn R. He explains some things that might require no explanation if you were learning R in the context of a statistics class.
The last four chapters would be interesting even for an experienced R programmer:
- Performance enhancement: memory and speed
- Interfacing R to other languages
- Parallel R
No one would be surprised to see the same chapters in a Java textbook if you replaced “R” with “Java” in the titles. But these topics are not typical in a book on R. They wouldn’t come up in a statistics class because they don’t provide any statistical functionality per se. As long as you don’t make mistakes, don’t care how long your code takes to run, and don’t need to interact with anything else, these chapters are unnecessary. But of course these chapters are quite necessary in practice.
As I mentioned up front, I haven’t read the book carefully. So I’m going out on a limb a little here, but I think this may be the book I’d recommend for someone wanting to learn R, especially for someone with more experience in programming than statistics.
Related post: R: The Good Parts