Statistical Modeling with Spline Functions

This monograph describes methodology, theory and applications of the use of polynomial splines in data mining. Over the last decade or so, the use of such splines has gained considerable popularity. This monograph will be the first book that discusses spline methods where both the location of the knots and the coefficients are optimized. After a preliminary chapter describing various properties of splines that are needed later on, the book discusses a number of well known methodologies and their variations in detail. These methodologies include MARS and POLYMARS (Chapter 3), POLYCLASS (Chapter 5), Logspline (Chapter 6), HARE (Chapter 7), Lspec (Chapter 8) and Triogram (Chapter 9). The last two chapters of the book give a thorough and comprehensive discussion of the theory behind polynomial spline methodologies. This monograph is aimed at statistical researchers and graduate students, as well as applied researchers using nonparametric statistical methods.