Bayesian Inference for the Multiple Linear Regression Model

This chapter develops the methods for fitting a linear regression model for the response variable y on a set of predictor variables from data comprising various points. The method of least squares is only a data analysis tool, which depends only on the data, not the probability distribution of the data. The Bears data, which can be found in both the Minitab example folder and the Bolstad package, contains a set of morphometric measurements as well as sex on a number of bears of various ages. The figures in the lower triangle of the matrix of the given in the chapter are the linear correlation coefficients for the pairs of variables. Statisticians can see that all of the continuous predictor variables have a moderately high correlation with weight.