Statistical modelling of road traffic accidents forms the original insight and implementation for road safety policies. Recently, there is an emerging trend to more progressively analysing road traffic accidents data using time series techniques. However, with sophisticated tools utilised for identification and optimally fitting time series models, it should be sensitive to inaccurate Automatic model building by the software. The purpose of this study is to examine automatic time series models for determination orders of parameters in the road traffic accidents data compared to the manual process by R software. Time series traffic accidents data were gathered for eighteen years from secondary sources, and statistical time-series analyses were performed. Time series decomposition, stationarity and seasonality were checked to identify the Introduction For decades, statistical analysis and modelling have been playing a significant role in getting insights from road traffic accidents data. Practitioners in this field encourage to use statistical modelling, analysis and forecasting to identify the root causes of problems and establish a foundation for evolving policies and economics based interventions (Zhang, Pang, Cui, Stallones, & Xiang, 2015). appropriate models for road traffic accidents. Meanwhile, automatically analysis of data was conducted with a comparison of both the results. AIC, BIC and other error values were used to choose the best models and model diagnostics tools were applied to confirm the statistical assumptions. SARIMA(0,1,2)(1,0,2)12 and SARIMA(0,1,2)(0,0,2)12 models resulted the best models manually and automatically, respectively. The diagnostic process showed that SARIMA (0,1,2)(1,0,2)12 performed better than the automatic model. Therefore, the modellers who prefer to use the Automatic function as a tool for time series model selection should consider the model's accuracy. It would be better for assessments to compare a variety of models and select the one having the best goodness of fit.

Statistical modelling of road traffic accidents forms the original insight and implementation for road safety policies. Recently, there is an emerging trend to more progressively analysing road traffic accidents data using time series techniques. However, with sophisticated tools utilised for ...