<![CDATA[The Industry 5.0 revolution influences business effectiveness by prioritizing digital transformation to unlock significant growth opportunities for MSMEs in Indonesia's culinary and handicraft sectors—which play a vital role in the national economy—particularly through the optimization of production and sales processes. A major challenge facing these MSMEs is product demand uncertainty. This study aims to improve demand prediction accuracy to enhance customer service and minimize inefficiencies associated with overstocking and understocking (which lead to lost sales opportunities). It explores the implementation of machine learning—specifically Support Vector Regression (SVR)—to predict product demand and boost MSME efficiency and competitiveness in a highly competitive digital era, analyzing key factors influencing business success. The dataset comprises 10 MSMEs, 20 MSME-product series, and 480 observations spanning January 2024 to December 2025. The model incorporates sales lags, moving averages, seasonal events, the rainy season, unit prices, and a 12-month cyclical component. The SVR model yielded an MAE of 19.54 units, an RMSE of 35.67 units, and a MAPE of 5.92%. While the model outperformed the seasonal-naive approach (MAPE 6.35%), it did not surpass the naive lag-1 method (MAPE 5.07%). However, the inclusion of external factors reduced the MAPE from 6.57% to 5.92% compared to an SVR model without such factors. These prediction results can serve as a foundation for production and inventory planning, with future developments intended to incorporate more dynamic pricing variables and comprehensive market factors.]]>
Copyrights © 2026
