ARIMA模型在发病率预测中的应用

Applications of ARIMA Model on Predictive Incidence

目的 探讨季节性时间序列ARIMA预测模型在时间序列资料分析中的应用 ,建立HFRS发病率的预测模型。方法 采用最大似然法估计模型参数 ,按照残差不相关原则、简洁原则确定模型的结构 ,依据AIC与BIC准则确定模型的阶数 ,建立ARIMA预测模型。结果 季节自回归参数有统计学意义 ,方差估计值为 4 2 3 0 ,AIC =3 0 9.5 2 3 ,BIC =3 11.78。对模型进行白噪声残差分析 ,χ2 检验表明ARIMA( 0 ,1,0 ) ( 1,0 ,0 ) 12 模型是适合的。结论 用所建模型对HFRS各月发病率进行了预测 。

Objective To discuss the application of seasonal time series ARIMA predictive model and fit predictive model of HFRS incidence. Methods Parameter of model is estimated based on maximum likelihood. The structure is determined according to criteria of residual un-correlation and concision. ARIMA predictive model was fitted and the order of model was confirmed through Akaike Information Criterion (AIC, for short) and Bayesian Information Criterion (BIC, for short). Results Statistics assisted estimation of the significance of the fitted seasonal Auto-regressive coefficients. The estimation of variance is 4.230, AIC=309.523,BIC=311.78. Using analysis of white-noise residual of model, ARIMA(0,1,0)(1,0,0) 12 seems to be the most appropriate model by χ 2 test. Conclusion The model of ARIMA can be used to forecast for HFRS incidence with high prediction precision of short-term time series.