基于残差修正的ARIMA-BP组合模型在中国戊型肝炎发病预测中的应用

Application of ARIMA-BP hybrid model on incidence prediction of hepatitis E in China based on residual correction

摘要:
目的探讨基于残差修正的ARIMA-BP组合模型在中国戊型肝炎传染病流行趋势预测中的作用。
方法对2004-2017年中国戊型肝炎统计数据采用SPSS软件分别建立ARIMA和ARIMA-BP模型，将2018年1-12月戊肝数据作为对比值，对模型的预测效果进行对比分析。
结果2种模型的预测结果评价指标中，ARIMA-BP组合模型的E、ER、MAE、MSE、MAPE指标整体上均小于ARIMA模型。
结论ARIMA-BP组合模型的预测效果优于ARIMA模型，可用于我国戊型肝炎发病趋势的早期预测。

Abstract:
ObjectiveTo explore the role of ARIMA-BP hybrid model based on residual correction in predicting the epidemic trend of hepatitis E infectious diseases in China.
MethodsThe ARIMA and ARIMA-BP models were established by SPSS software based on the statistical data of hepatitis E in China from January 2004 to December 2017.The data of hepatitis E in China from January 2018 to December 2018 were used as the comparison value to analyze the prediction effect of the model.
ResultsThe overall index values of E, ER, MAE, MSE and MAPE of ARIMA-BP hybrid model were smaller than those of ARIMA model.
ConclusionsThe prediction effect of ARIMA-BP hybrid model is better than that of ARIMA model, which can be used for the early prediction of the incidence trend of hepatitis E in China.