This study examines the application of various machine learning regression models to predict pollutant concentrations for evaluating air quality. Air pollution poses a significant global health hazard, and accurate air quality index (AQI) prediction is critical for effective environmental management and public health policy. The research utilizes air quality data from the World Health Organization (WHO) Ambient Air Quality Database and employs a robust data preparation pipeline. A comparison of linear regression and nonlinear regression is presented. Quantitative results show that the Random Forest Regressor achieves the highest predictive accuracy with an \(R^2\) score of 0.812 and the lowest Mean Squared Error (MSE) of 20.23, outperforming other models significantly. The K-Nearest Neighbors and Decision Tree Regressors also demonstrate strong performance, with \(R^2\) scores of 0.776 and 0.762, respectively. In contrast, linear models exhibit very low \(R^2\) values (around 0.16) and high MSEs (above 90), indicating their limited ability to capture nonlinear patterns in the data. These findings demonstrate the superiority of ensemble methods, particularly Random Forest, in modeling complex environmental relationships and highlight the importance of using advanced regression techniques for reliable air quality forecasting.