HYBRID ML MODELS FOR ACCURATE AIR POLLUTION FORECASTING
Keywords:
Hybrid Machine Learning , Air Pollution Forecasting , Artificial Neural Networks (ANN), Support Vector Machines (SVM), Long Short-Term Memory (LSTM)Abstract
A hybrid machine learning architecture that includes many prediction methods improves air pollution model reliability and efficiency in this paper. This allows more accurate forecasts. The paper uses ANNs, SVMs, and LSTM networks to find linear and curved air quality patterns. Meteorological data, temporal correlations, and historical pollution levels help the hybrid model predict more accurately than separate models. Ensemble approaches improve prediction accuracy and dependability, while feature selection and normalization improve input data quality. The hybrid method's improved RMSE and MAE performance suggests it could be used for real-time air quality monitoring and environmental decision-making.