Browse Books

Reviewer: Daniel M. Berry

The title suggests that this book describes only an object-oriented environment for statistical computing and dynamic graphics. Given my long-standing aversion to things statistical and probabilistic, when I read the title, I said to myself, “Oh, how dreary!” I was tempted to return the book, saying that I would not be a good reviewer, but good citizenship got the better of me and I decided to weather the book. I was surprised—t he book turned out to be well written and fun to read. It contains a nice tutorial on LISP that is, in fact, the clearest I have read in a long time, maybe because the focus is on using LISP. The explanations are clear and to the point and are accompanied by well-chosen examples. The book also contains a good tutorial on object-oriented programming. It covers the standard topics of objects and inheritance, both single and multiple, again strictly from the point of view of using them. Then the book develops windows, menus, dialogue, graphics windows, statistical graphics windows, and dynamic statistical graphics windows (note the inheritance implicit in this list), all as a carefully constructed hierarchy of objects. Because of the above-mentioned aversion, I am in no position to judge the book from the viewpoint of the practicing statistician. The book is valuable as a software engineering text that gives examples of the object-oriented construction of a real, used tool, however. A real plus from this point of view is that the software described in the book, XLISP-STAT, is available by anonymous ftp for a variety of computing platforms. An instructor who wishes to use the book as a resource can make a running version available for experimentation. I can see the book and the software being used in a software engineering class to allow students to see how a complete system is put together in an object-oriented manner; course exercises could include building enhancements as subclasses.