Aaron Esdon
Assessing the Contributing Factors in Serious Traffic Collisions for Durham Region Using Log-Linear Models and Spatial Analysis © 2004
The recently available 2003 Traffic Collision dataset from Durham Regional Police was utilized to develop a series of models and analysis designed to identify hotspots of serious motor vehicle collisions causing injury. In the analysis, prior claims and hypotheses were addressed to determine if the contributing factors behind serious traffic collisions were being given appropriate attention. The resulting paper focused on a number of environmental, urban, and spatial characteristics, and their relationship to collisions causing injury.
It has been an overriding belief that traffic volume is directly related to the presence of serious traffic collisions. This claim was addressed, and the results indicate that hotspots of traffic volume are not positively related to collisions causing injury. Due to these results, other variables present in the dataset were examined, some of which were temporal in nature. Day of the week, and month of the year variables were utilized in a series of chi-squared tests to determine if the expected temporal distribution was significantly different than the distribution present in the dataset.
For the most part, the forecasting of serious traffic collisions causing injury was determined using a series of general log-linear models. Explanatory variables were selected through the construction of contingency tables, and the examination of chi-squared measures. Consequently, the interaction between environmental and light conditions was discovered to be valuable in predicting traffic collision incidents. Additionally, traffic signal variables, impact type and their relationship to traffic collisions were also determined to be related. Nevertheless, further analysis utilizing 2004 DRPS traffic collision data will be required for examination of the explanatory variables and their importance in predicting traffic collision hotspots.