<p>The optimization of machine learning models to air quality index (AQI) prediction through hyperparameter tuning is the main emphasis of this study. To help with health risk awareness and regulatory actions, AQIs are essential for evaluating and informing the public about air quality. To improve AQI predictions, machine learning models such as Decision Tree (DT), Random Forest (RF), eXtreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine (LightGBM), and Multilayer Perceptron (MLP) have been used in the literature. Hyperparameters have a major impact on these models performance and must be carefully adjusted for best outcomes. In this paper, DT, RF, XGBoost, LightGBM, and MLP were fine-tuned using a variety of hyperparameter Optimization approaches, including Grid Search Cross-Validation (GridSearchCV), Random Search Cross-Validation (RandomSearchCV), Simulated Annealing (SA), and Genetic Algorithms (GA). The article provides a thorough analysis of different techniques, highlighting their effects on model efficiency and forecast accuracy. The results provide insightful information about the efficacy of various machine learning models and hyperparameter tuning strategies in the context of air quality forecasting. According to experimental findings, ensemble models such as XGBoost and LightGBM fared better than others in terms of computational efficiency and prediction accuracy. As compared to grid-based searches, evolutionary tuning techniques like GA and SA produced better parameter sets faster. The study concludes that XGBoost provides the best overall results for AQI forecasting when tweaked via GA, and that appropriate hyperparameter optimization greatly enhances model performance. These results offer important information for creating reliable prediction models that will aid in environmental monitoring and policy formation.</p>