ABSTRACT
Nonparametric methods play two distinct, but related, roles in the exploratory analysis of spatial point process data. First, they are used to test benchmark hypotheses about the underlying process. For example, most analyses of univariate spatial point process data begin with one or more tests of the hypothesis of complete spatial randomness (CSR), by which we mean that the data form a partial realization of a homogeneous Poisson process. Although this is rarely tenable as a scientific hypothesis, formally testing it serves to establish whether the data contain sufficient information to justify any more elaborate form of analysis. The second role for nonparametric methods is to estimate properties of the underlying process with a view to suggesting suitable classes of parametric model.
