Air quality prediction-based big data analytics using hebbian concordance and attention-based long short-term memory.

Area of Science:

• Environmental Science
• Data Science
• Public Health

Background:

• Dwindling air quality due to economic development poses significant health risks, particularly from Particulate Matter (PM) 2.5.
• Accurate prediction of PM 2.5 concentrations is essential for public health protection.
• The complex nature of air pollution necessitates advanced prediction methodologies.

Purpose of the Study:

• To propose a novel method, Hebbian Concordance and Attention-based Long Short-Term Memory (HC-ALSTM), for accurate air quality prediction.
• To address the complexity of air quality prediction by integrating advanced feature extraction and selection techniques.
• To improve the accuracy and efficiency of predicting Particulate Matter (PM) 2.5 concentrations.

Main Methods:

• Preprocessing raw data using Statistical Normalization to manage pollutant impacts.
• Extracting dimensionality-reduced spatio-temporal features with the Generalised Hebbian Spatio Temporal Feature extraction algorithm.
• Selecting significant features using Concordance Correlation analysis to identify pollutant influences.
• Employing Attention-based Long Short-Term Memory for precise air quality prediction.

Main Results:

• The HC-ALSTM method demonstrated superior performance in terms of prediction accuracy and computational time compared to existing methods.
• Feature selection using Concordance Correlation effectively identified key pollutant impacts for prediction.
• The model achieved significant improvements in air quality prediction accuracy and overhead.

Conclusions:

• The proposed HC-ALSTM method offers a highly effective approach for accurate air quality prediction.
• This method provides a valuable tool for mitigating the adverse health effects of air pollution.
• HC-ALSTM represents a substantial advancement in spatio-temporal air quality forecasting.