Jüt tüketim tahmini: Halı imalatinda bir vaka çalışması

In today's increasingly competitive market conditions, it is necessary for the all companies to meet the demand of the customers in a timely and complete manner. For this reason, all resources (raw materials, semi-finished products, all energy sources, etc.) should be prepared and supplied at the right time and in sufficient quantities. This situation forces companies to make forecasts of raw materials to avoid possible lateness. The most basic forecasting methods are statistical estimation methods. However, these statistical methods may be inadequate in the system with high uncertainties and high dynamism. At this point, the method of artificial neural network (ANN), which have been succesful and widespread in recent years, steps in. The studies show that the results obtained from ANNs may be much more successful than the statistical methods applied in different areas for estimation. Jute is one of the most basic raw materials used in carpet production and it is important to estimate the jute consumption correctly in order to prevent possible production delays and to ensure customer satisfaction. In this study, jute consumption estimations were made by ANN method using data for the years between 2015-2017 of carpet company. In this thesis, a multiple linear regression (MLR) model was also established to evaluate the performance of ANN method. The results show that ANN is more successful than regression analysis considering the performance indicators' values (values of MSE- mean squared error- R2 ve adjusted R2).