Overcoming data shortages through hybrid machine learning in air quality forecasting: a case study of Kağıthane, Istanbul
Theoretical and Applied Climatology, cilt.157, sa.10, 2026 (SCI-Expanded, Scopus)
- Yayın Türü: Makale / Tam Makale
- Cilt numarası: 157 Sayı: 10
- Basım Tarihi: 2026
- Doi Numarası: 10.1007/s00704-026-06545-9
- Dergi Adı: Theoretical and Applied Climatology
- Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, IBZ Online, Aerospace Database, Artic & Antarctic Regions, BIOSIS, Environment Index, Geobase, Index Islamicus, INSPEC, Academic Search Ultimate (EBSCO), Natural Science Collection (ProQuest), Biomedical Reference Collection: Corporate Edition (EBSCO), Earth, Atmospheric, & Aquatic Science Collection (ProQuest), Materials Science & Engineering Collection (ProQuest), Technology Collection (ProQuest)
- Akdeniz Üniversitesi Adresli: Evet
Özet
Accurate prediction of the Air Quality Index (AQI) is essential for protecting public health and supporting evidence-based environmental policy, particularly in rapidly urbanizing regions where air-quality monitoring networks are limited. This study proposes a hybrid machine-learning framework for AQI prediction in Kağıthane, Istanbul, Türkiye, using a long-term dataset spanning approximately ten years, thereby reducing reliance on dense sensor infrastructures. The methodology consists of two stages. First, K-Means clustering is applied to meteorological variables and pollutant concentrations to identify distinct atmospheric regimes associated with characteristic pollution patterns. Second, XGBoost regression is employed to estimate AQI values using meteorological parameters and key pollutants, including PM₁₀, PM₂.₅, SO₂, CO, and NO₂. By leveraging widely available meteorological data, the proposed framework mitigates data scarcity while maintaining high predictive capability. The model achieves an overall accuracy of approximately 98%, enabling timely health advisories and more efficient allocation of air-quality management resources. Feature-importance analysis further identifies the dominant meteorological drivers governing AQI variability, providing actionable insights for urban planners and decision-makers. Owing to its reliance on commonly available datasets and its modular design, the framework is readily transferable across different geographic and temporal contexts, offering a scalable solution for air-quality assessment in regions facing environmental monitoring constraints.