Examining spatial relationships between crashes and the built environment: A geographically weighted regression approach

Research output: Contribution to journalArticlepeer-review

28 Scopus citations

Abstract

A better understanding of the relationships between vehicle crashes and the built environment is an important step in improving crash prediction and providing sound policy recommendations that could reduce the occurrence or severity of crashes. Global statistical models are widely used to explore the relationships between vehicle crashes and the built environment, but these models do not incorporate a spatial component and are unable to deal with the issues of spatial autocorrelation and spatial non-stationarity. Our research utilizes a geographically weighted regression (GWR) model to explore the relationships between crashes and the built environment in the context of the Detroit region in Michigan. We find that the relationships between the built environment and crashes are spatially non-stationary: both the strength and the direction of their relationships differ over space. Our study also identifies several built environment variables, such as commercial use percentage, local road mileage percentage, and intersection density, that have relatively stable relationships with crashes. Our research demonstrates the feasibility and value of using spatial models in traffic, transportation, and land use research.

Original languageEnglish
Pages (from-to)221-233
Number of pages13
JournalJournal of Transport Geography
Volume69
DOIs
StatePublished - May 2018

Keywords

  • Built environment
  • Detroit region
  • Geographically weighted regression
  • Spatial non-stationarity
  • Vehicle crashes

Fingerprint

Dive into the research topics of 'Examining spatial relationships between crashes and the built environment: A geographically weighted regression approach'. Together they form a unique fingerprint.

Cite this