Model Prediksi Konsentrasi PM2.5 di Wilayah Jakarta Menggunakan Algoritma Random Forest

Arsy, Muhammad Naufal Afif Al
Yasir, Ahmad Meijlan

Journal of Computation Physics and Earth Science
Vol 5 No 1 (2025): Journal of Computation Physics and Earth Science

This study predicts PM2.5 concentrations in Jakarta using the Random Forest algorithm with historical air quality data from 2015 to 2024. Hyperparameter tuning was performed to optimize model performance, focusing on parameters such as n_estimators, max_depth, and min_samples_split. The model achieved a Mean Absolute Error (MAE) of 14.44, a Root Mean Square Error (RMSE) of 18.75, and an R² Score of 0.61. While the model captured general PM2.5 fluctuation patterns, deviations at certain points indicate room for improvement. Descriptive analysis showed an average PM2.5 concentration of 94.46 µg/m³, with peaks up to 209 µg/m³, exceeding healthy air quality thresholds. The model can be integrated into real-time monitoring systems and support data-driven policies. Future work could incorporate meteorological variables and evaluate longer-term trends to enhance accuracy.