Using adaptive neural-fuzzy inference systems for demand forecasting and application with comparison artificial neural network method

Due to the rapid increase in global competition among organizations and companies, rational approaches in decision making have become indispensable for organizations in today?s world. Establishing a safe and robust path through uncertainties and risks depends on the decision units? ability of using scientific methods as well as technology. Demand forecasting is known to be one of the most critical problems in organizations. A company which supports its demand forecasting mechanism with scientific methodologies could increase its productivity and efficiency in all other functions.Artificial neural networks are frequently being used as a decision-making mechanism in organizations and companies recently. In addition, organizations often use applications integrated with fuzzy logic methodologies to model and solve their problems that comprise uncertainty.In this study, it is aimed to solve a critical demand forecasting problem with artificial neural networks and fuzzy logic methods. In the first phase of the study, the factors which impact demand forecasting are determined, and then a database of the model is established using these factors. The parameters and their optimal values used within artificial neural networks and ANFIS are derived by several trials and tests. The results estimated by each of these two methods are comparatively analyzed and also these results are compared with the results obtained by traditional methods.Keywords: Fuzzy Logic, Artificial Neural Networks, Demand Forecasting, Anfis