A Hybrid Model for Forecasting Sales in Turkish Paint Industry

Abstract

Sales forecasting is important for facilitating effective and efficient allocation of scarce resources. However, how to best model and forecast sales has been a long-standing issue. There is no best forecasting method that is applicable in all circumstances. Therefore, confidence in the accuracy of sales forecasts is achieved by corroborating the results using two or more methods. This paper proposes a hybrid forecasting model that uses an artificial intelligence method (AI) with multiple linear regression (MLR) to predict product sales for the largest Turkish paint producer. In the hybrid model, three different AI methods, fuzzy rule-based system (FRBS), artificial neural network (ANN) and adaptive neuro fuzzy network (ANFIS), are used and compared to each other. The results indicate that FRBS yields better forecasting accuracy in terms of root mean squared error (RMSE) and mean absolute percentage error (MAPE).