ProfARIMA: A profit-driven order identification algorithm for ARIMA models in sales forecasting

Tine Van Calster, Bart Baesens, Wilfried Lemahieu

Abstract: Highlights•ProfARIMA is a new order identification algorithm for Seasonal ARIMA models in sales forecasting.•It implements a flexible profit function as the fitness function of an evolutionary algorithm.•The algorithm is applied to publicly available datasets and datasets from The Coca-Cola Company.•In most cases, ProfARIMA selects a more profitable model than the traditional Box-Jenkins method.