Optimized PM2.5 Prediction Model Using XGBoost Coupled with Improved Grasshopper Optimization Algorithm

Accurate prediction of PM2.5 concentrations is required to curb the risks of air pollution on human health. This study applies an approach to an extended gradient boosting model, advanced to an integrated improved grasshopper optimization algorithm, to predict PM2.5, using both meteorological and chemical inputs. The advantages of XGBoost are utilized to manage obscure linkages and nonlinear interactions, whereas IGOA has the advantage of effectively optimizing hyperparameters (HPs). Therefore, we also conducted a comparative analysis between the following six regression models: linear regression (LR), random forest (RF), decision tree (DT), gradient boosting (GBoost), extended gradient boosting (XGBoost), and XGBoost_GOA with the proposed XGBoost_IGOA. The performance of the model was analyzed using various sets of input variables. According to the experimental data, the proposed model yields appreciable improvements in PM2.5, with an MSE (μg/m3) of 216, MAE (μg/m3) of 5, MAPE (%) of 15.34, and R2 of 0.812. The results of this study suggest that the proposed method is a valuable tool for predicting PM2.5 concentration accurately, and it may help other researchers in the execution of an improved prediction model in their respective areas.