Chapter 15 Nonparametric Spatial Autocorrelation

In this chapter, the spatial weights matrix is abandoned in the formulation of a spatial autocorrelation statistic. Instead, a measure of attribute (dis)similarity is related directly to the distance that separates pairs of observations, in a non-parametric fashion. Two approaches covered are the spatial correlogram and the smoothed distance scatter plot.

To illustrate these methods, I will use the Italy Community Banks point layer sample data set.