<![CDATA[Accurate sales forecasting is a component in inventory management, particularly for determining appropriate stock levels for upcoming periods. Forecasting inaccuracies may result in overstocking or understocking, leading to increased operational costs and decreased service quality to customers. In practice, many retail businesses still encounter difficulties in producing reliable sales forecasts due to fluctuating demand patterns and the use of forecasting methods with suboptimal parameter selection. Single Exponential Smoothing (SES) is widely used because of its simplicity and ease of implementation; however, its forecasting performance is highly dependent on the choice of the smoothing parameter (alpha), which is often determined using a trial-and-error approach. This study proposes a hybrid forecasting approach that combines Single Exponential Smoothing with Golden Section Search to enhance sales prediction accuracy. Golden Section Search is employed as a numerical optimization technique to systematically determine the optimal alpha value by minimizing forecasting errors measured using Mean Absolute Percentage Error (MAPE). The proposed approach is applied to sales data from XYZ as a case study using varying lengths of historical data, namely 3 months, 6 months, 12 months, 24 months, and 36 months. The results demonstrate that the proposed hybrid method is capable of producing forecasts with a good level of accuracy, particularly for short-term forecasting. The lowest MAPE value of 7.02% is achieved when using 3 months of historical data, indicating high responsiveness to recent demand changes. As the length of historical data increases, the model tends to become more stable but less responsive to trend fluctuations, resulting in higher error values. Overall, the proposed approach is effective in supporting inventory management decision-making by providing accurate and reliable sales forecasts.]]>
