中圖分類號： TP391.9
文獻標識碼： A
DOI： 10.19358/j.issn.2097-1788.2022.06.012
引用格式： 徐海峰，黃小莉，張政. 基于自適應Boosting組合模型的空氣質量預測[J].網絡安全與數據治理，2022，41(6)：84-89.

Air quality prediction based on adaptive boosting combinatorial model

Abstract： Aiming at the large error of the current air quality prediction model and the fact that a single model has certain limitations in different aspects, resulting in large prediction error, an adaptive boosting combination model is proposed. Five methods, including the squared error sum reciprocal method and the simple weighted average method, are used to adaptively assign weights to the three basic boosting models. The final result is superimposed according to the corresponding weights, to give full play to the advantages of each single model and improve the prediction accuracy. The experimental results show that, the combination of squared error and reciprocal method performs optimally, and the weighted combinatorial model using the error squared and reciprocal method was calculated with MAE 7.124 4, RMSE 9.367 1, and R2 as 0.863 9，better than other weighting combination methods and a single boosting model. The application of this combined model provides an effective method for optimizing air quality prediction systems.

Key words : air quality；adaptive；boosting model composition model；error squared and reciprocal method