The phenomenon of air pollution is a problematic and pressing global concern as it has greatly affected human health and the environment. The appropriate formulation of various kinds of models predicting air quality index (AQI) will help to avoid health hazards due to poor air quality and take preemptive actions. This contribution discusses the use of ML algorithms to predict the AQI by leveraging historical air data with relevant weather conditions. In this work, we carry out an experimental comprehensive study involving several ML algorithms such as linear regression, decision trees, support vector machine, and ensemble methods applied in making predictions from the AQI model. A few popular measures: mean absolute error (MAE) was evaluated along with its root mean square error (RMSE) as well as R-squared, wherein the performances of different models have been found to be quantified. The study indicates that incorporating ensemble methods into the modeling of AQI is most effective as compared to conventional models suggestive of the efficiency of ML approach in the field of air quality monitoring. For KNN, the accuracy is 80%, and for random forest, the accuracy is 83%.