<![CDATA[The imbalance in raw material inventory due to fluctuations in demand for fast-moving menu items is a major challenge for restaurants in operational efficiency. This study aims to determine the most accurate forecasting method to predict demand for nine fast-moving menu items at the Pagi Sore Pemuda Restaurant to optimize inventory control. Using monthly sales transaction data from April 2025 to February 2026, the study applies four forecasting methods: Prophet, Holt-Winters, EWT-Prophet, and EWT-Holt-Winters with a time series forecasting design. Accuracy is measured using the Mean Absolute Percentage Error (MAPE) with a training data division (the first nine months) and testing (the last two months). The results show that the EWT-Prophet method provides the best performance with an average MAPE of 14.02%, outperforming the standard Prophet (15.53%), EWT-Holt-Winters (30.75%), and Holt-Winters (90.61%). Eight menu items achieved good to excellent accuracy (MAPE<20%), with Beef Minced Curry being the most accurate (9.25%) and Mashed Sweet Potato being fairly accurate (21.22%). The three-month forecast (March-May 2026) identified a downward trend across all menu items compared to the previous period, ranging from 6.3% to 27.2%. The three main menu items (Ayam Pop, Rendang, Gulai Tunjang) remained dominant, contributing more than 80% of total sales. These findings provide a quantitative basis for management to adjust the volume of raw material purchases per menu, reduce food waste, and improve overall inventory control efficiency.]]>
