Linear discriminant analysis

This chapter gives a first view of how we can interactively explore a fitted model, in the context of the parametric linear discriminant analysis. It starts by explaining which elements of the model are important, and how they can be extracted from a fitted model. It then shows how we can check the model assumptions visually, first for two dimensional examples, and then extending to higher dimensions using the tour. A next section then describes how we can examine the model fit, by comparing the class boundaries from the fitted model to the data. It also shows how this is done in higher dimensions by using a slice tour display. Finally, the interpretation of a fitted linear discriminant analysis is discussed, in particular in the context of how a local tour can be used to understand the uniqueness of the discriminant space for a clear separation between groups.