The effect of Operation 24 Hours on reducing collision in the City of Edmonton

In the City of Edmonton, in order to reduce the prevalence of collisions, the Operation 24 Hours program (OPS24) was developed by using existing police and transportation services resources. The program uses traditional manned police speed enforcement method, which are supplemented by traffic safety messages displayed on permanent and mobile dynamic messaging signs (DMS). In this paper, collision data analysis was performed by looking at the daily number of collisions from 2008 to 2011 that covers 28 Operation 24 Hours (OPS24) events. The objective of the collision data analysis is to analyze if there is a reduction in collision frequencies after OPS24 was held and examined how long the collision reduction effect last. Weather factors such as temperature, thickness of snow, and wind gust have been considered by many as a great influence on collision occurrences, especially in a city with long and cold winter such as Edmonton. Therefore, collision modeling was performed by considering these external weather factors. To analyze the linear and periodic trend of different collision types (injury, fatal, and property damage only (PDO)) and examine the influence of weather factors on collisions, negative binomial time series model that accounts for seasonality and weather factors was used to model daily collision data. The modeling also considered collision proportion to account for missing traffic volume data; the Gaussian time series model that accounts for seasonality and weather factors was used to model collision proportion. To estimate the collision trend and test for changes in collision levels before/after OPS24, interrupted time series model with segmented regression was used. While for estimating how long the effect of the OPS24 last, change point method was applied.