Abstract

Using GWR To Predict the Number of Crashes at Intersections
Track: Transportation
Authors: LIXIN HUANG

Geographically Weighted Regression (GWR) and Ordinary Least Squares (OLS) were used to develop the model predicting the number of crashes at intersections. Annual Average Daily Traffic (AADT) data of major and minor roadway were selected as the explanatory variables. Based on the availability of traffic volume data, a total of 63 intersections in Miami-Dade County Florida were identified to develop the model. The total number of crashes per year for the roadway was selected as the dependent variable. The crash data of five years were obtained for the study area. Five models were developed on yearly bases. The model results from GWR and OLS were compared to each other for each year. It was found that spatial variations occurred in the model developed. It means that the relationship between the number of crashes and AADT varied over space. Therefore, GWR is the appropriate method to develop the model.