GM(1,1)灰色模型和ARIMA模型在HFRS发病率预测中的比较研究

目的对GM(1,1)模型和ARIMA模型在肾综合征出血热(HFRS)发病率预测中的效果进行比较。方法利用1990-2001年辽宁省、丹东市和沈阳市HFRS的发病率分别建立GM(1,1)灰色预测模型和ARIMA模型,对建立的模型进行拟合。同时,对2002年3个地区的HFRS发病率进行预测,比较2个模型的拟合和预测效果。结果针对辽宁省HFRS发病率建立的GM(1,1)模型和ARIMA模型的平均误差率(MER)分别为13.5143%、25.0814%;决定系数(R2)分别为0.8961、0.6997。针对丹东市HFRS发病率建立模型的MER分别为19.7329%、20.6275%;R2分别为0.8112、0.7628。针对沈阳市HFRS发病率建立模型的MER分别为15.1421%、18.0584%;R2分别为0.8757、0.7889。结论GM(1,1)模型对于小样本以及隐含指数函数变化趋势的资料具有明显的预测优势,预测效果优于ARIMA模型,对解决时间序列类型的HFRS发病率等资料有很好的实用价值。

Comparison of GM (1,1) Gray Model and ARIMA Model in Forecasting the Incidence of Hemorrhagic Fever with Renal Syndrome

Objective To compare the effects of GM(1,1) model and ARIMA model in forecasting the incidence of hemorrhagic fever with reual syndrome(HFRS). Methods GM(1,1) model and ARIMA model were established with the HFRS incidence of Liaoning province,Dandong and Shenyang from 1990 to 2001 and the two models were fit together. The forecast was made to the HFRS incidence in the three areas in 2002,and the effects of fitting and forecasting were compared. Results The MER of GM(1,1) model and ARIMA model for Liaoning province was 13.5143% and 25.0814%,and the R2 of the two models 0.8961 and 0.6997 respectively. The MER of GM(1,1) model and ARIMA model for Dandong was 19.7329% and 20.6275%,and the R2 of the two models 0.8112 and 0.7628 respectively. The MER of GM(1,1) model and ARIMA model for Shenyang was 15.1421% and 18.0584%,and the R2 of the two models 0.8757 and 0.7889 respectively. Conclusion GM(1,1) model is superior in small samples on forecasting,and its effect is better than that of ARIMA model,so it is of practical value in dealing with time series data such as the incidence of HFRS .