The increase in vehicle and industrial pollution in air is rising. Air pollution is the major factor driven behind the environmental pollution. An increase in air pollution leads to chronic lung diseases and lower respiratory. In order to check the pollution in time, it is necessary to know about it. Using historical data and machine leaning models, it becomes easier to track the daily increase in air pollution levels. This will help to restrict the air pollution. This paper focuses on the prediction of different pollutants affecting human lungs. Different regression algorithms have been used to predict the amount of particulate matter (PM2.5) in the air which leads to the pollution. This paper focuses on linear regression, ridge regression, lasso regression, support vector regression, and gradient boosting regression for analyzing and predicting air pollution levels. Gradient boosting regression has better accuracy as compared to other regression techniques. Cross-validation is used to ensure the data does not have any outliers. The different regression algorithms have been evaluated using different parameters like root mean squared error (RMSE), R-squared score, mean absolute error (MAE), and mean squared error (MSE).