Air pollution resulting from vehicular emissions in urban areas is known to cause breathing discomfort and respiratory illnesses in people who are exposed to such emissions. In this paper we have proposed a multi-objective machine learning model that can classify air quality based on presence of pollutants, predict the quality of air in a specified region based on past temporal data and suggest actions to control exposure of individuals to severely polluted areas. Our pollution model considers the most common pollutants affecting air quality in urban areas which includes NO2, SO2, CO, PM2.5, PM10, NH3 and Ozone. We have experimented with multiple supervised learning models including Artificial Neural Networks, Multi-class Support Vector Machines and Naive Bayesian Model. Based on our experiments, we propose using a naive Bayesian model for prediction with a corresponding set of rules that indicate actions to be taken when pollution reaches or is predicted to reach breach levels.