An Enhanced Framework for Forecasting Marine Accidents with Seasonal and Casualty Severity Categories as Exogenous Factors

Maritime accident prediction is an important aspect of safeguarding maritime management including insurance companies. Data related to maritime accidents have characteristics that are more complex with the nature of both non-stationarity and non-linearity. Thus predicting maritime accidents is a challenging task in the field of research. In this paper, we have developed a new framework considering the Seasonal Autoregressive Integrated Moving Average with Exogenous Variables (SARI-MAX) model trained with Casualty Severity feature having three categories Very Serious Marine Casualty, Marine Casualty, and Marine Incident as exogenous variables to forecast the amount of global maritime accidents. Experimental results show that our proposed model provides an R-squared value of 0.999 which depicts almost accurate and robust predictions. Our framework provides a systematic way to perform time series forecasting by optimizing the SARIMAX model including both seasonal and exogenous factors for better accuracy. The proposed framework provides effective decision support to maritime professionals to do risk management with appropriate resource allocation.