ABSTRACT
This chapter generalizes a regression scenario in several ways and considers the multiple regression setting where the mean of a continuous response is written as a function of several predictor variables. It describes the methodology for comparing different regression models and considers the case where the response variable is binary with two possible responses. The chapter explains modeling of the probability of a particular response as a function of a predictor variable. A general method of comparing models is called cross-validation. In this method, one partitions the dataset into two parts: the training and testing components. One initially fits each regression model to the training dataset. Then one uses each fitted model to predict the response variable in the testing dataset.
