On the Modeling of Traffic Accident Risk in Nganjuk Regency by Poisson Point Process on a Linear Network

Methods for analyzing point patterns in two-dimensional space are well developed. However, in reality, an object or event can be limited to occur along a network of lines called a linear network, for example, a traffic accident which occurs within a road. The aims of this study are to derive the parameter estimation procedure for the log-linear Poisson point process model on a linear network and to apply the model to analyze the distribution of traffic accident locations in Nganjuk Regency involving two covariates, namely the road type and the distance of accident location to the closest traffic light. Parameter estimation is carried out by applying the Berman-Turner scheme to the log-likelihood function, which is then maximized using the Iteratively Reweighted Least Square (IRLS) method. The results of the analysis show that the two covariates are significant to determine the risk of a traffic accident in Nganjuk, but based on the AIC value, the best model obtained is the model with covariate the distance of accident location to the closest traffic light. We expect more risk on a primary arterial road. On the road with a distance 1 meter closer to a traffic light, the risk for a traffic accident occurrence increases 1.001 times.