The visual analysis of spatial regression

Thomas William Pederson, University of Pennsylvania


The primary objective of this research is to develop a set of digital visualization techniques with which to examine residual autocorrelation in spatial regression models using a spatial weight matrix. Its secondary objective is to explore and refine the process by which these techniques are developed as a prototype for the more general design of user-centered, task-driven components in a geographic information system. The research performed visual experiments on the results of spatial regression models that informed the design of an prototypical tool set developed for a specific category of spatial statistical analysis. The approach is presented as a paradigm for the design of other such tools. It includes user-centered, task-driven components that consists of a collection of spatial statistical and visualization techniques developed to facilitate the examination of residual autocorrelation using a spatial weight matrix. The research resulted in an interface design that combines ease of use, intuition, and the appeal of visualization with the relative certainty of statistics.

Subject Area

Urban planning|Area planning & development

Recommended Citation

Pederson, Thomas William, "The visual analysis of spatial regression" (2000). Dissertations available from ProQuest. AAI9965540.