摘要
目的探讨传递函数模型在我国5岁以下儿童死亡率(child mortality rate under age 5,U5MR)预测中的应用价值。方法利用1991至2004年国家有关统计年鉴数据拟合传递函数模型,同时采用求和自回归移动平均(ARIMA)模型进行预测,并比较两者的拟合结果。结果传递函数模型与实测值比较而言,其回代预测误差为-1.11%,未来预测误差为-2.35%,均较ARIMA模型低。结论传递函数模型在预测中可以纳入有关影响因素,该预测结果理论上更符合实际情况。
Objective To explore the applicability of transfer function-noise model in the prediction of child mortality rate under age 5（U5MR）. Methods Transfer function-noise model and auto regressive integrated moving average （ARIMA） model were used to fit the officially released data and compare the prediction results. Results The prediction error of transfer function-noise model is -2.35%, which is lower than that of ARIMA model. Conclusion The related factors can be taken into account and the prediction results are more precise by fitting the transfer function-noise model.
出处
《中华妇幼临床医学杂志（电子版）》
CAS
2008年第2期25-27,共3页
Chinese Journal of Obstetrics & Gynecology and Pediatrics(Electronic Edition)
关键词
传递函数模型
ARIMA模型
5岁以下儿童死亡率
预测
transfer function-noise model
ARIMA model
child mortality rate under age 5 （U5MR）
prediction
