Analisis Performa Metode Extreme Learning Machine dan Multiple Linear Regression dalam Prediksi Produksi Gula

Sugar is a crucial commodity in Indonesia, with demand increasing annually. Variations in sugar production require accurate prediction strategies for industrial planning. This study aims to analyze the performance of the Extreme Learning Machine (ELM) and Multiple Linear Regression (MLR) methods in predicting sugar production. This research employs a quantitative experimental approach, with sugar production data during the 2020-2023 milling period as the research subject. Data collection techniques involve observation and documentation, while data analysis techniques utilize Mean Absolute Percentage Error (MAPE) and 10-Fold Cross-Validation to measure model accuracy. The results indicate that ELM has a lower error rate (MAPE 16.06%) compared to MLR (MAPE 27.90%), making it more effective in capturing complex sugar production patterns. Implementing this model in a web-based system also enables more efficient production monitoring. The ELM method proves to be superior in predicting sugar production and can be integrated into industrial systems to support data-driven decision-making. Future research can explore other predictive models, such as deep learning, and consider external factors like weather and soil conditions to enhance accuracy.