温特斯法和季节指数GM（1，1）模型在狂犬病暴露人群预测中的应用  

Comparison of winters model and seasonal-index GM（1,1 ） model in forecasting population exposed to rabies

作　　者：顾全忠[1] 任飞林[1] 

机构地区：[1]浙江省舟山市疾病预防控制中心传染病防制科,316021

出　　处：《国际流行病学传染病学杂志》2013年第6期397-401,共5页International Journal of Epidemiology and Infectious Disease

摘　　要：目的比较温特斯法模型和季节指数GM（1，1）模型对舟山市狂犬病暴露人群数量的预测效果。方法选取2006—2011年舟山市各月狂犬病暴露人群资料分别建立温特斯法模型和季节指数GM（1，1）模型，用2012年1—12月实际暴露人数验证比较两种模型拟合预测效果。选取相对误差较小的模型预测2013年舟山市狂犬病暴露人数。结果应用季节指数GM（1，1）模型和温特斯法模型对2006--2011年狂犬病暴露人群资料进行拟合，模型平均绝对误差率分别为19．1048％和7．7013％；温特斯法模型的平均绝对误差、均方根误差都低于季节指数GM（1，1）模型。2012年季节指数GM（1，1）模型和温特斯法模型预测值距实测值的平均绝对误差率分别22．8435％和11．3124％。温斯特法模型预测2013年舟山市狂犬病暴露约13526人。结论温特斯法模型对舟山市狂犬病暴露人群的预测效果要优于季节指数GM（1，1）模型。在预测具有季节性和趋势的资料时，应用两种模型进行比较后选择最优模型，更有利于疾病的预测和防治效果的评估。Objective To compare the prediction effect of winters model and the seasonal-index GM （1,1） model for the number of population exposed to rabies in Zhoushan city. Methods The seasonal-index GM（1,1） model and winters model were established on the monthly populations exposed to rabies in Zhoushan from 2006 to 2011. The prediction effect of the two models were compared and verified by the real data in 2012. The model with the less relative error was adopted to predict the population in Zhoushan in 2013. Results The seasonal-index GM （1,1） model and winters model Were used to analyze the date of population exposed to rabies from 2006 to 2011. The mean absolute percent error were 19.104 8%and 7.701 3% respectively. The average absolute error and root mean square error of the winters model were both lower than those of seasonal-index GM （1,1） model. The mean absolute percent error of winters model and seasonal-index GM （1,1） model were 11.312 4%and 22.843 5% respectively.The number of the people who exposed to rabies, which forecasted would be 13 526 in Zhoushan in 2013 by the winters model. Conclusions The winters model is superior to seasonal-index GM （1,1） model in predicting populations exposed to rabies in Zhoushan. The two models should be compared to choose a better model to predict the seasonal or tendency data, and is beneficial to disease prediction and assessment of the control effect.

关 键 词：狂犬病 GM(1 1) 温特斯法 季节性 

分 类 号：R512.99[医药卫生—内科学]