Air quality prediction is critical for public health and environmental management. The study employs machine learning techniques to forecast air quality indices based on historical data and environmental variables. The dataset comprising meteorological conditions, pollutant levels, and temporal features to train various models, including linear regression, decision trees. This paper presents a hybrid classification model combining Random Forest and Naive Bayesian is proposed for air quality classification. Mutual Information-based feature selection is used to select informative features. The model achieves improved accuracy and robustness. The hybrid approach leverages strengths of both classifiers. Effective air quality classification is achieved. The experimental results conducted on Air quality India dataset with performance metrics such as Root Mean Square Error and R2 were evaluated to assess model accuracy.