Urban Air Quality Classification Using Machine Learning Approach to Enhance Environmental Monitoring

Abstract

Urban areas worldwide grapple with environmental challenges, notably air pollution. DKI Jakarta, Indonesia's capital city, is emblematic of this struggle, where rapid urbanization contributes to increased pollutants. This study employed the CatBoost machine learning algorithm, known for its resistance to overfitting and capability to handle missing data, to predict urban air quality based on pollutant levels from 2010 to 2021. The dataset, sourced from Jakarta's air quality monitoring stations, includes pollutants such as PM10, SO2, CO, O3, and NO2. After preprocessing, we used 80% of the data for training and 20% for testing. The model displayed high accuracy (0.9781), precision (0.9722), and recall (0.9728). The feature importance chart revealed O3 (Ozone) as the top influencer of air quality predictions, followed by PM10. Our findings highlight the dominant pollutants affecting urban air quality in Jakarta, Indonesia and emphasizing the need for targeted strategies to reduce their concentrations and ensure a cleaner and healthier urban environment.