Demand forecasting in a manufacturing business with artificial neural networks and ANFİS method

Since the sales of our company in our study are generally in European countries and various other countries, while the sales of one country decrease, the sales of the other country increase. Examining these factors can increase the accuracy of the models used in demand forecasting using methods such as artificial neural networks (ANN) and ANFIS and is thought to contribute to the strategic planning of the business. The data used in the study are the actual amounts sold by our company for 2 years. The data was taken from SAP Business One, an ERP (Enterprise Resource Planning) program. The use of both ANN and ANFIS methods in this study covers many important reasons such as comparing model performance, determining strengths and weaknesses, performing more comprehensive analysis, examining different application areas, providing academic and practical contributions and obtaining more reliable predictions. With this approach, it is aimed to increase the scientific value of the thesis study and maximize the prediction accuracy. As a result, it has been observed that both models can make successful predictions on certain datasets and that their performance varies depending on the training ratio. The ANFIS model generally provides higher prediction accuracy and stability, while the ANN model can deliver effective results under specific conditions. This study demonstrates that both ANFIS and ANN are valuable methods for time series forecasting and can complement each other in certain situations. These findings suggest that both methods can be effectively used for time series forecasting and that the choice of method may depend on the structure of the dataset.