Ship collision frequency prediction with Bayesian spatiotemporal log-logistic model

The Bayesian hierarchical log-logistic model is introduced to handle the incompleteness of the observation in space-time effects so as to improve the accuracy in analyzing the quantitative relation between the frequency of ship collision and the influencing factors（the speed and magnitude of traffic flow, traffic density, waterway width and other environmental factors）. The method is used to process the AIS data from the Yangtze river estuary in January-September 2014. The research shows that The Bayesian hierarchy log-logistic model with ability of handling unknown spatiotemporal effects has better fitness than ordinary log-logistic model does in terms of deviance information criterion. It also indicates that the traffic density is the most significant risk factor, and collision risk is considerably higher in area with higher traffic density. Weather and time（day or night） also have great impact on the risk.