Daily PM_2.5 concentration prediction based on principal component analysis and LSSVM optimized by cuckoo search...

Area of Science:

• Environmental Science
• Data Science
• Machine Learning

Background:

• Particulate Matter (PM2.5) pollution poses significant environmental and health risks, particularly in China.
• Accurate PM2.5 concentration forecasting is crucial for effective monitoring and control strategies.
• Existing models may lack the precision required for comprehensive air quality management.

Purpose of the Study:

• To develop a novel hybrid model for precise daily PM2.5 concentration forecasting.
• To enhance prediction accuracy by integrating feature reduction and intelligent parameter optimization.
• To evaluate the proposed model's performance against established forecasting methods.

Main Methods:

• Principal Component Analysis (PCA) for feature extraction and dimensionality reduction.
• Least Squares Support Vector Machine (LSSVM) for PM2.5 concentration prediction.
• Cuckoo Search (CS) algorithm for optimizing LSSVM hyperparameters to improve generalization.

Main Results:

• The proposed PCA-LSSVM-CS hybrid model demonstrated superior performance in PM2.5 prediction.
• The model outperformed a standard LSSVM with default parameters.
• Comparative analysis showed better accuracy than a General Regression Neural Network (GRNN) model.

Conclusions:

• The hybrid PCA-LSSVM-CS model offers a highly precise approach for PM2.5 forecasting.
• This model has significant potential for application in real-world air quality forecasting systems.
• The study highlights the effectiveness of combining PCA, LSSVM, and CS for environmental data prediction.