Air pollution has been an important research topic because of the detrimental consequences that PM2.5 has on human health and environmental sustainability. This study assesses how well machine learning algorithms predict the Air Quality Index (AQI) and looks into how PM2.5 concentrations affect the AQI. To measure the correlation between PM2.5 and AQI, a dataset comprising 2191 observations of PM2.5 and AQI values was examined. According to a linear regression model, the AQI rises by 0.92 units for every 1 µg/m3 increase in PM2.5, indicating a high positive correlation (r = 0.99) between the two variables. With a R2 value of 0.96, the model showed that 96% of the variation in AQI can be explained by PM2.5. Four machine learning techniques were also assessed for AQI prediction: Support Vector Machine (SVM), Random Forest, Decision Tree, and Linear Regression. Decision Tree and Random Forest achieved lowest RMSE and the highest R2, while SVM obtained the lowest MAE. The Prophet model also showed satisfactory results for forecasting. These results demonstrate impact of PM2.5 in determining air quality and role of machine learning models prediction and forecasting.