Forecasting the Number of Firemen Interventions Using Exponential Smoothing Methods: A Case Study

Abstract

Predicting the number of firemen interventions to size the appropriate workload of firefighters to the appropriate need is vital for reducing material and human resources. Therefore, it will have a great impact on reducing the financial crisis resulting from global warming and population growth. The database in this research includes interventions recorded hourly from “1 January, 2015 00:00:00” to “31 December, 2019 23:00:00” in Doubs, France. The data were processed, decomposed, outliers were detected and replaced. Thenceforth, optimal smoothing values were selected and then three different models of Exponential Smoothing were deployed. Experiments have shown that Holt-Winters’ method has the best accuracy comparing to the baseline and other Exponential Smoothing techniques. The results are promising and would optimize the number of firefighters’ resources.