Air Quality Analysis and Prediction Based on the Adaboost-VAR Model

This paper is based on nearly one and a half years of air quality data in China, with a focus on the air quality of Hebei Province. Using grey correlation analysis, PM2.5, PM10, SO2, NO2, O3, CO were taken as the feature data, and AQI was taken as the label data. Adaboost model was constructed to analyze the weight value of each feature vector. PM10 and PM2.5 have the greatest influence on AQI. Adaboost model has high precision and can better calculate AQI through six eigenvectors. Then the six groups of feature data are processed by simple moving average and ADF test, which shows that there is no unit root in the series, and the VAR vector autoregressive model can be constructed. After determining the lag order, the trend for the next 30 days is then predicted for each of the six sets of eigenvectors. Finally, the prediction results of are used as the characteristic data of Adaboost to predict the future AQI value, and the air quality level of the region is divided. The application of the model aims to provide more accurate air quality forecasts by utilizing historical data and capturing complex temporal patterns of pollutant concentrations. The model can help policy makers and environmental agencies make informed decisions, set realistic targets, and implement timely interventions to mitigate pollution.