Performance of machine learning models to forecast PM10 levels

Area of Science:

• Environmental Science
• Data Science
• Atmospheric Chemistry

Background:

• Particulate Matter 10 (PM10) poses significant environmental and health risks.
• Accurate forecasting of PM10 concentrations is crucial for public health and environmental management.
• Machine learning offers advanced capabilities for complex environmental data analysis.

Purpose of the Study:

• To develop and evaluate an optimized machine learning methodology for forecasting PM10 concentrations at predefined locations.
• To compare the performance of eight different machine learning models for PM10 prediction.
• To identify the most effective model for accurate, location-specific PM10 forecasting.

Main Methods:

• A comparative analysis of eight machine learning models was conducted.
• An ensemble model integrating state-of-the-art techniques was developed.
• The models considered air quality and meteorological factors for forecasting.

Main Results:

• The ensemble model significantly outperformed the other seven models.
• The developed methodology achieved a high regression coefficient (R²≈1) across all tested models.
• The study confirmed the effectiveness of machine learning for location-specific environmental factor prediction.

Conclusions:

• Machine learning, particularly ensemble methods, provides a powerful tool for accurate PM10 forecasting.
• The case-specific methodology enhances the precision of environmental predictions.
• This approach has potential for broader applications in predicting location-specific environmental factors.