Home Statistical Dictionary Spatial Point Process
Spatial Point Process
A point process specifically defined on a spatial domain (typically the plane or three-dimensional space), used to model the random locations of objects like trees, disease cases, or crime incidents, with the Poisson point process as its simplest, no-interaction baseline case.
In Plain English
A spatial point process is a point process where the underlying space is literal, physical space, a map, a plot of land, a volume, rather than time or some abstract index. This is the framework statistical ecologists and epidemiologists use to ask questions like, are these tree locations clustered together, spread out evenly, or scattered with no pattern at all? The Poisson point process (complete spatial randomness, points scattered with no interaction) is the reference baseline every spatial point pattern gets compared against; genuinely interesting spatial point processes usually build in either clustering (points attracting nearby points) or inhibition (points repelling nearby points).
Definition
A spatial point process is a point process defined on a spatial domain, typically or , describing the random locations of a set of points (events, objects, individuals). It is the specifically spatial member of the general point process family, and is characterized by its intensity function (the expected density of points per unit area/volume at location ) and, for describing interaction between points, its pair correlation function or K-function. The Poisson point process, with constant intensity and fully independent point locations, is the null model of complete spatial randomness that spatial statistics tests against. Beyond this baseline, spatial point processes are broadly classified as clustered (points aggregate, as modeled by a Cox process or Neyman-Scott process), or regular/inhibited (points spread out more evenly than random, as modeled by a Strauss process or hard-core process). Analytical tools including Ripley's K function, quadrat counts, and nearest-neighbor distance tests are used to detect and characterize which regime an observed spatial point pattern falls into.
Properties
- This category's Point Process entry gives the fully general framework (any underlying space); this entry is specifically the spatial-domain case, and shares its detection toolkit (Ripley's K function, quadrat counts, nearest-neighbor tests) directly with the Spatial Statistics dictionary category, since 'spatial point process' is the theoretical umbrella that concept lives under.
- Ecology, epidemiology, and criminology are the classic application domains: tree locations in a forest plot, disease case locations relative to a suspected source, and crime incident locations across a city are all textbook spatial point pattern datasets analyzed with this exact framework.
- The choice between clustered and regular/inhibited models isn't just a statistical fitting exercise, it usually reflects a real scientific hypothesis, e.g. disease clustering suggesting a common environmental cause, or plant inhibition reflecting competition for resources, making spatial point process modeling a direct link between raw location data and substantive interpretation.
At a Glance