Application of Support Vector Regression and Random Forest Methods in Predicting Air Quality Index (AQI) Concentration in Jakarta

Air pollution is a significant global challenge that profoundly impacts both the environment and public health. Accurate prediction of the Air Quality Index (AQI) is essential for promoting awareness and effectively managing pollution. This study systematically compares the performance of Support Vector Regression (SVR) and Random Forest (RF) models in forecasting Jakarta's AQI, utilizing daily data collected from January 1, 2018, to December 31, 2023. The models were thoroughly evaluated using key performance metrics, including Adjusted R-squared, Mean Absolute Error (MAE), and Mean Absolute Percentage Error (MAPE). The dataset was divided into three distinct periods: pre-COVID, during COVID, and post-COVID. The results clearly indicate that both models performed exceptionally well during the pre-COVID period. The SVR model achieved an impressive Adjusted R-squared of 96.37%, alongside a MAE of 4.09 and a MAPE of 5.63% using two combinations of independent variables. The RF model closely followed, attaining an Adjusted R-squared of 95.49%, a MAE of 4.48, and a MAPE of 6.16% based on seven combinations of independent variables. These findings affirm the effectiveness of both models in accurately predicting Jakarta's AQI. However, the comparative analysis underscores that the SVR model slightly outperforms the RF model in terms of precision and accuracy, particularly during the pre-COVID period. Practical applications of these models include the development of real-time AQI monitoring platforms, simulation of policy impacts, event-specific strategies, and integration into long-term urban planning to mitigate air pollution and protect public health.