I got the following review of Object-oriented Programming in R on Amazon:
I am, of course, not happy with only three stars, but I don’t think it is entirely wrong. I don’t think I give any good examples of the usefulness of object-oriented programming in R, and of the three R programming books, this, I feel, is the weakest. I do explain the various object-oriented systems in R, but do not really explain why object-oriented programming is any use in R.
For day-to-day analysis work in R, you typically use a lot of object-oriented features, but you don’t program them yourself. There, you are more likely to use functional programming. I had a hard time trying to come up with good exampels of object-oriented programming you would actually use in R, which delayed this book a lot, and I never really succeeded.
The examples I have with adding meta-information to numbers to add physical units, I think are good examples of how object-oriented systems can be used, but the truth is that I mostly use classes when implementing new statistical models, and I didn’t include that in the book. I had some plans for writing about that in a separate book, but I am not sure there is enough to write about there for a full book, so maybe I should add a chapter to this book about it instead.
When you fit different models to data and use the generic methods like predict() or coefficients() for analysis, you use the object-oriented features of R. Since I want to make this book a little longer before I can sell it as a reasonably length paperback book to Apress, such a chapter could kill two birds with one stone. I’ll have a look at this over the next week.