Prediction of Air Quality Index Using Machine Learning Techniques: A Comparative Analysis

An index for reporting air quality is called the (AQI). It measures impact of pollution on a person’s health over short period time. The purpose AQI to educate public negative effects local pollution. amount in Indian cities has significantly increased. There are several ways create mathematical formula determine index. Numerous studies have found link between exposure and adverse impacts population. Data mining techniques one most interesting approaches forecast analyze it. aim this paper find effective way prediction assist climate control. method can be improved upon optimal solution. Hence, work involves intensive research addition novel such as SMOTE make sure that best possible solution problem obtained. Another important goal demonstrate display exact metrics involved our it educational insightful hence provides proper comparisons assists future researchers. In proposed work, three distinct methods—support vector regression (SVR), random forest (RFR), CatBoost (CR)—have been utilized New Delhi, Bangalore, Kolkata, Hyderabad. After comparing results imbalanced datasets, was lowest root mean square error (RMSE) values Bangalore (0.5674), Kolkata (0.1403), Hyderabad (0.3826), well higher accuracy compared SVR (90.9700%) (78.3672%), while RMSE value Delhi (0.2792) highest obtained (79.8622%) (68.6860%). Regarding dataset subjected synthetic minority oversampling technique (SMOTE) algorithm, noted (0.0988) (0.0628) accuracies (93.7438%) (97.6080%) comparison regression, whereas (85.0847%) (90.3071%). This demonstrated definitely datasets had algorithm applied them produced accuracy. novelty lies fact models picked through thorough by analyzing their accuracies. Moreover, unlike related papers, balancing carried out SMOTE. all implementations documented via graphs metrics, which clearly show contrast help what actually caused improvement