Nonparametric Statistical Methods Using R (Chapman & Hall/CRC Texts in Statistical Science) Kloke, John and McKean, Joseph

It is difficult to find a good book to teach a course on nonparametric statistical methods. You need to have a good statistical background to understand this book, and it is information dense, reading more like a reference manual. I found it appealing that it relied heavily on R, but I wish there were more exercises it were more readable for students. The approach is modern, and the authors promote their procedures over older ones. I will stick with this book when I teach the course again if I cannot find a simpler one that is easier to read.

I have known Joe McKean (and John Kloke) for many years. His work covers the implementation and application of nonparametric and robust methods based on well developed theory. There are many situations where classical statistical methods produce incorrect results when analyzing large real data. I knew of these issues even before Big Data became a buzz word. They were difficult because they could appear without any diagnostic indication of an underlying problem. In addition to working in HPC, I have been an avid R language user since S was first released, using it in both computer science classes and many industrial consulting projects. I now almost never drop back to SAS or other systems, even while working on FDA submissions.We have just created a new Data Science major, joint between statistics and computer science, at Western Michigan University. Much of the work in this new major will require learning R. There are two relevant courses within the major. The first is an advanced R programming course. The second is a course in managing and analyzing Big Data stored on distributed heterogeneous systems, Hadoop and other systems, as well as streaming data. It is my view that even if a classic LS analysis is going to be reported, a corresponding robust nonparametric analysis should be done for confirmation. This is important for data sets of all sizes.This book provides a practical and accessible introduction to using nonparametrics in R. It requires some basic background knowledge in statistics and existing high quality open source code. One of the book's strengths is that it contains further references to both theory and applications, pointing the way for students to do further research beyond the text. The authors have also contributed very usable code in this area to the R CRAN archives, some of which is referenced in the text itself.I highly recommend this book.

Met my expectations.