Forecasting and decision support system development for continuous production system: A sample application for chemical sector

Companies need to analyze their demands for strategic decisions in the future. The most important input in demand forecast analysis is historical data. Accurate analysis of historical demand data will help enterprises make the right decision. Enterprises producing high value-added products in the chemical sector must have an accurate demand forecasting system in order to manage the supply chain with optimum resource utilization. In this study, demand forecasting using time veries analysis methods was applied to products in the AX group according to result that products, were analyzed by ABC-XYZ classification method, of a company in the chemical sector that not having a quantitative demand forecasting system, has a high raw material stock level, has a tendency to be out of stocks and excess stock in the final products. 32 different products were first classified into ABC classes and then XYZ classification was made considering the coefficient of variation. 3 periods demand forecasting for products in the A and X group was applied by ARIMA and Holt-Winter methods using 96 monthly data between 2011 Jan. and 2018 Dec.. ARIMA and Holt Winter forecasting results were compared with MAPE, MAD and RMSE error criterias. Considering to up to %8,1 better performance of Holt-Winter method, a decision support system has been developed that optimizes the parameters to provide the lowest MAPE value.