Vector generalized linear and additive models. With an implementation in R (Q500275)

In this book, a much larger and more flexible statistical framework is presented that has greatly expanded generalized linear models for regression modeling, which centers on vector generalized linear models (VGLMs), vector generalized additive models (VGAMs), and their variants with implementation in R. Within this framework, statistical models presented in the book are related to each other in a natural way, which thus provides a natural vehicle for thinking about regression modeling, and performing much more of applied statistics as a coherent whole. The book consists of two parts containing 18 chapters and an appendix. Part I contains 9 chapters, which describes the general theory and computational details behind each major class of models. Chapter 1 gives a brief overview of the entire framework. Chapter 2 reviews linear models, generalized linear models, univariate smoothing methods, and generalized additive models. Chapter 3 describes VGLMs, which lie at the heart of the overall statistical framework, being the linear or parametric class. Chapter 4 presents VGAMs, which are VGLMs based on smoothing. Chapters 5 introduces reduced-rank VGLMs (RR-VGLMs), an extension of the VGLMs, whose primary defining characteristic is that they operate on latent variables. In Chapter 6, RR-VGLMs are extended to give the quadratic RR-VGLMs, which enables one to perform constrained quadratic ordination. Chapter 7 describes the class of RR-VGLMs, which performs constrained additive ordination. Chapter 8 introduces the VGAM package. Chapter 9 addresses a few topics not conveniently discussed in previous chapters. Part II explores some major application areas. Chapter 10 surveys a few miscellaneous models that may be considered extensions of linear models and binomial generalized linear models, and which are accommodated with in the VGLM/VGAM framework. Chapters 11--12 summarize a collection of univariate discrete and continuous distributions that are presented as VGLSs/VGAMs, and have been implemented as VGAM family functions. Chapter 13 considers bivariate continuous distributions including bivariate normal distribution, Plackett's bivariate distribution, and a special class of bivariate distributions called copulas. Chapter 14 describes VGAM family functions for a categorical response variable. Chapter 15 is devoted to quantile and expectile regression. Chapter 16 focuses on extremes. Chapter 17 centers on zero-inflated, zero-altered and positive discrete distributions. Chapter 18 gives some details about what a VGAM family function consists of. Background material is given in the appendix. Exercises are provided in the end of each chapter. In addition, R code is integrated in the text, and datasets are used throughout. This book can serve as a textbook for senior undergraduate or first-year postgraduate courses on generalized linear models or categorical data analysis. This book is also an excellent resource for statisticians, applied statisticians, natural scientists and social scientists.