Evidence record · UDINUS · SDGs
- Unit
- Rektorat UDINUS
- Period
- 2024
- Goals
- Record
- 27c83be8-493a-40ad-ace2-51648c30eb74
- Published
- 2026-10-30
Universitas Dian Nuswantoro conducted an activity to develop an accurate air quality prediction model for addressing air pollution challenges in Jakarta. The activity involved Wibowo, Wahyu; Al Azies, Harun; Wilujeng, Susi A.; and Abdul-Rahman, Shuzlina as participants. Data from conventional air pollution indexes recorded by air quality monitoring stations was employed. During the research phase, the data was explored and SMOTE was used to manage imbalances. XGBoost was then used to develop a model with the best parameters.
The evaluation stage showed the model's ability to predict air quality. The reported accuracy rate was 99.516%, the F1-score was 99.528%, and the recall rate was 99.509%. These performance indicators showed the model's ability to classify and predict air quality levels. The activity also investigated the significance of various variables in predicting air quality. The evaluation considered weight, gain, total gain, and cover indicators. The findings indicated that numerous aspects were significant for prediction. SO2 helped predict air quality, while the prevalence of PM2.5 on several measures revealed a significant influence.
Through this research activity, Universitas Dian Nuswantoro applied data exploration, imbalance management, XGBoost modelling, and variable-significance evaluation to an air quality challenge in Jakarta. The activity contributes knowledge about the complicated dynamics of air quality prediction and provides information relevant to addressing air pollution issues and promoting healthier urban environments.