Optimizing The Ordering Policy By Forecasting Demand And Control Inventory

Abstract

The primary goal of this thesis is to formulate effective methodologies for forecasting intermittent demand in the raw material sector of an outdoor company. In the pursuit of identifying the most efficient approach, a thorough examination of three distinct strategies was undertaken, ranging from fundamental techniques to models specifically designed for predicting intermittent demand items. SES, a strategy for short-term forecasting, operates under the assumption of a relatively stable mean in the data without any discernible trend. The Moving Average (MA) method aims to predict demand by calculating an average quantity of actual demand from previous periods. Notable for its adaptability and versatility in predictive analytics, linear regression emerged as a valuable technique. The selection of the optimal method was based on its capacity to significantly reduce errors compared to actual demand. Throughout the research, there was a focus on precision and accuracy, aiming to improve the predictability of spare parts within the outdoor furniture industry. To determine the optimal value, a conventional optimization strategy was employed, considering the lowest Mean Absolute Percentage Error (MAPE), Mean Absolute Error (MAE), and Root Mean Square Error (RMSE). Furthermore, for cost optimization, this study incorporates inventory management techniques such as ABC analysis, Reorder Point (ROP), Economic Order Quantity (EOQ), and Safety Stock into the forecasting approach.