ABSTRACT
This chapter presents S functions and objects for classical linear methods in statistics, in which a numerical response variable is predicted by linear combinations of other numeric or categorical variables. The statistical, computational, and mathematical ground deserves to be called "classic" in any sense. The statistical use of linear models goes back to Laplace and Gauss early in the nineteenth century and continues to underlie much of statistical modeling. Even such a well-developed area of statistics as linear models provides many challenges and opportunities. New functions, new user interfaces, and new algorithms can all be built on the basis provided. A wide range of plots and summaries can be applied to linear models. After fitting a model to some data, we would often like to know the predicted response from the model for some different values of the predictor variables.
