ARMA模型对肾综合征出血热发病趋势预测的拟合研究  

FITTING RESEARCH ON ARMA MODE IN THE PREDICTION OF INCIDENCE TREND OF HEMORRHAGIC FEVER WITH RENAL SYNDROME

作　　者：陈叶[1] 白杉[2] 陈慧中[1] 孙百军[2] 魏文婧[1] 黄牧[1] 王萍[2] 

机构地区：[1]沈阳市疾病预防控制中心,沈阳110031 [2]沈阳市预防医学会

出　　处：《现代预防医学》2008年第8期1414-1415,共2页Modern Preventive Medicine

摘　　要：[目的]探讨应用ARMA模型在时间序列资料分析中的应用,建立HFRS发病趋势的预测模型,从理论角度为制定科学、有效的防制措施提供理论依据。[方法]应用平稳时间序列分析,选用最小二乘法按照信息量准则,AIC、模型的拟合度与相关系数以及误差,确定ARMA的所有参数,用非线性最小二乘法重新估计,得出组合模型的最终估计;应用DPS(Data Processing System)数据处理系统。[结果]数据转换后,建立ARMA(1,1)模型,其残差平方和SS=0.6987,残差标准差=0.2158,AIC=-3.7044,相关系数R=0.96930,拟合度C=93.96%。[结论]模型可以用来进行预测。[Objective] To explore the application of ARMA model in the analysis of time series data, to establish the prediction model for predicting incidence trend of the hemorrhagic fever with renal syndrome （HFRS）, and to provide theoretical basis for formulating scientific and effective preventive and control measures based on the aspect of theory. [ Methods] The parameters of ARMA were determined by using stationary time series analysis, least square method in line with the information content, AIC, the fitting of model, coefficient correlation and relative accuracy. The parameters of ARMA were evaluated again by using non-linear least square method to get the final estimate value of combination model; The Data processing system （DPS） were performed to analyze these data we collected. [ Results] The MRMA model was established after data transformation. The sum of squares of residues was 0.698 7 and the standard deviation of residual error was 0.215 8, AIC was 3.704 4, coefficient correlation was 0.969 30 as well as the fitting index was 93.96%. [Conclusion] The model could use to prediction.

关 键 词：ARMA模型 时间序列 发病率 

分 类 号：R195.1[医药卫生—卫生统计学] R512.8[医药卫生—卫生事业管理]