Discriminant Analysis and Statistical Pattern Recognition (Wiley Series in Probability and Statistics)

Rating:
Summary: scholarly and thorough treatment of discriminant analysis
Review: Geoff McLachlan has written a thorough and up-to-date text on discriminant analysis and pattern recognition. There are a number of fine books on discriminant analysis. McLachlan's is one of the best. It is at an intermediate level and provides many of the recent advances including regularized discriminant analysis.

Discriminant analysis and pattern recognition are very similar topics. The term discriminant analysis is common in the statistical literature while pattern recognition is more common in the electrical engineering literature. McLachlan is scholarly and familiar with the literature in both disciplines (not common). He includes over 1200 references with many references from the late 1980s.

Professor McLachlan has been a key contributor to the literature on error rate estimation in discriminant analysis and devotes a great deal of coverage to this important topic. He also includes recent developments on bootstrap methods and summarizes the literature on bootstrap methods for adjusting bias in error rate estimation.

Much of the bootstrap work on error rate estimation involves comparative simulation studies, particular when training sample sizes are small. McLachlan provides a nice summary of the his work, the work of Efron and he also includes discussion of a couple of my simulation studies co-authored with Murthy and Nealy.