Abstract
Often, industries are confronted with incomplete information for forecasting decisions. Unfortunately, very few models can serve this purpose. In addressing this gap, a univariate model for random and fluctuating fire accidents is presented. A data turbulence monitoring mechanism – information turbulence index (ITI) – and an out-of-sample analysis-based forecasting tool – the disjoint pair-point exponential and time-weighted-average (DPEWTA) model – characterise the model. Comparative analysis of the model's performance against that of other forecasting models was investigated. Simulated outcomes indicate DPEWTA's relative superior and inferior forecast capabilities up to 22 and 5%, respectively, over ARIMA, ESM and MA at low-to-medium ITI values. Real system DPEWTA forecasts show MAPE and MAE values with the best prediction ranges obtained for an ITI of 0.20 to 0.25. Obtained results show the model's capability and dependability as a fire accident forecasting tool.
Disclosure statement
No potential conflict of interest was reported by the authors.