ABSTRACT
We now present a development of exploratory tools and modeling approaches that are customarily applied to data collected for areal units. Again, this literature is sometimes referred to as discrete spatial modeling to reflect the fact that we are only specifying a joint model for a finite set of random variables. We have in mind general, possibly irregular geographic units, but of course include the special case of regular grids of cells (pixels). Indeed, many of the ensuing models have been proposed for regular lattices of points and parameters, and sometimes even for point-referenced data (see Chapter 12 on the problem of inverting very large matrices).
