ABSTRACT
This chapter introduces and defines the conditional distribution model, and explains why it is the correct model for regression applications. It also defines the classical regression model in terms of the more general conditional distribution model, and proves that this model is incorrect because of the constraints it places on the more general conditional distribution model. Along the way, generalization is defined, the population terminology is eschewed in favor of the process terminology, the random-X viewpoint is defined and defended, and simulation is introduced as the key to easily understand all concepts presented in this book. Simple graphical and statistical analyses are performed using statistical software R on example data sets. LOESS is introduced as an estimate of the mean function. Finally, the origin of the term regression is explained; and the bivariate normal distribution is introduced along the way.
