Estimation of urban AQI based on interpretable machine learning.

Area of Science:

• Environmental Science
• Data Science
• Atmospheric Chemistry

Background:

• Air pollution poses a significant global challenge, impacting human health and quality of life.
• Accurate air quality prediction is crucial for effective pollution control and mitigation strategies.
• The Air Quality Index (AQI) provides a standardized measure for communicating air quality levels.

Purpose of the Study:

• To develop and evaluate machine learning models for estimating the Air Quality Index (AQI) in Shijiazhuang, China.
• To compare the performance of eXtreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine (LightGBM), and Random Forest (RF) models in AQI prediction.
• To identify key factors influencing AQI variations using model interpretability techniques.

Main Methods:

• Utilized machine learning algorithms including XGBoost, LightGBM, and RF for AQI estimation.
• Employed pollutant concentrations and meteorological factors as input variables for the models.
• Applied SHAP (SHapley Additive exPlanations) for model interpretation to determine feature importance.

Main Results:

• The XGBoost model achieved the highest prediction accuracy with an R² of 0.929, outperforming RF and LightGBM.
• PM2.5 and PM10 were identified as the primary contributors to AQI variations.
• Meteorological factors showed a less significant impact on AQI compared to particulate matter.

Conclusions:

• Machine learning approaches, particularly XGBoost, are effective for accurate air quality prediction.
• Understanding the influence of specific pollutants like PM2.5 and PM10 is vital for targeted air pollution management.
• The developed model demonstrates generalizability and good performance across different Chinese cities.