Prediksi Nilai Unburned Carbon Batubara yang Dihasilkan PLTU Menggunakan Algoritma Linear Regression, Random Forest, dan LightGBM Regression

Syarif, Muhyiddin
Afiyati

Jurnal JTIK (Jurnal Teknologi Informasi dan Komunikasi)
Vol 9 No 2 (2025): APRIL-JUNE 2025

This study focuses on predicting unburned carbon levels in coal-fired power plants to enhance operational efficiency. Accurate prediction of unburned carbon is crucial as it directly affects fuel combustion efficiency and environmental sustainability. The research compares three machine learning algorithms: Linear Regression, Random Forest, and LightGBM Regression, using performance metrics such as Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE), and Root Mean Squared Error (RMSE). The results show that LightGBM Regression performs the best, with MAE of 0.31, MAPE of 1.29, and RMSE of 0.38, outperforming the other two models. This model can be further optimized to improve prediction accuracy, contributing to more efficient and environmentally friendly power plant operations. The application of machine learning in this study supports data-driven decision-making in the energy sector.