Research On Air Quality Prediction Based On Model Fusion

As the environmental condition on which human beings live,air quality has attracted more and more attention from people.Therefore,both the state and individuals are making contributions to protect the environment.In order to put forward more effective protection measures,various factors affecting air quality should be fully analyzed and air quality should be predicted,so as to provide a good judgment basis for people to protect the environment and have some effects on people’s travel and life.In this paper,the hourly air pollution data set in changping district of Beijing was selected to evaluate the ARIMA model,BP neural network model and XGBoost model with eight evaluation indexes.The results show that XGBoost model performs best,followed by BP neural network model.In order to further improve the prediction accuracy,this paper uses five model fusion methods to fuse the above three models.In order to evaluate the effect of the mixed model,this paper chooses three general and specific indicators: root mean square error,adjusted R square and average absolute percentage error.The results show that three indicators of the fusion model are the best among the eight models.Therefore,this paper finally concluded that the fusion model was more accurate than the previous three single models when the XGBoost model,ARIMA model and BP model were fused with the method of reciprocal of sum of squares about prediction errors.