The objective of predictive examination is to forecast future trends by using cur-rent events as a basis. Regression, feature extraction, and data pre-processing are all steps in this process. Numerous characteristics and occurrences are present in the dataset used for the prediction analysis. The legitimate source is where the da-tastes are downloaded. The dataset is pre-processed to eliminate variables that are redundant or have multiple missing values. To determine which attribute has the most influence on the target set, the alliance between several impute and the goal is created in the second phase. Elastic net, Random Forest tree regression, LASSO, and linear regression are familiar with forecast air quality. In cases where the quantity of attributes is substantial, the air quality cannot be accurately predicted using regression models. A hybrid of straight, springy net, and random forest devolution is operated estimate air quality more accurately. The suggested model yields better results than other regression models already in use, with RMSE and R2 coming in at 1.8 and 0.92 respectively.