A Hypothesis Test for Detecting Spatial Patterns in Categorical Areal Data

Area of Science:

• Spatial Statistics
• Geographic Information Science
• Environmental Science

Background:

• The proliferation of spatial datasets necessitates advanced statistical methods for pattern detection.
• Spatial patterns like clustering and dispersion are key areas of research in areal data analysis.
• Existing methods often struggle with categorical variables and distinguishing nuanced spatial arrangements.

Purpose of the Study:

• To develop a novel hypothesis test for detecting spatial clustering or dispersion in categorical areal data.
• To extend the positive area proportion function to handle multi-category spatial variables.
• To enable the differentiation of various spatial patterns, including homogeneous and heterogeneous clusters, and dispersion.

Main Methods:

• Development of the categorical positive area proportion function test.
• Extension of existing methods for binary areal data to categorical data.
• Validation through an extensive simulation study.

Main Results:

• The proposed test effectively detects spatial clustering and dispersion in categorical areal data.
• The method successfully distinguishes between homogeneous clusters, heterogeneous clusters, and dispersion.
• The first approach capable of differentiating various types of clustering in categorical areal data has been established.

Conclusions:

• The categorical positive area proportion function test is a valuable tool for analyzing spatial patterns in categorical areal data.
• This method provides a significant advancement in understanding complex spatial arrangements.
• The test was successfully applied to analyze spatial patterns in land use data for Boulder County, Colorado.