乘积季节ARIMA模型在食源性疾病预测中的应用

Multiplicative Seasonal Autoregressive Integrated Moving Average Model Applied to Forecast Foodborne Diseases

目的探讨应用乘积季节自回归求和移动平均模型(autoregressive integrated moving average,ARIMA)预测如东县食源性疾病发病的可行性,为食源性疾病的预防和控制提供依据。方法基于2004年1月至2010年12月食源性疾病人数建立乘积季节ARIMA模型,用2011年食源性疾病资料验证模型的预测效果,用所得模型预测2012年食源性疾病发病人数。结果 ARIMA(0,1,1)×(0,1,1)12较好地拟合了既往时间段食源性疾病发病人数的时间序列,拟合平均相对误差为2.7%,预测2012年如东县食源性疾病发病总人数为64人。结论乘积季节ARIMA模型可以较好地拟合食源性疾病的时间变化趋势,并用于预测未来的食源性疾病,是一种短期预测精度较高的预测模型。

Objective To explore the feasibility for application of autoregressive integrated moving average （ARIMA） model to predict incidence of foodbome diseases in Rudong County and to provide scientific basis for the prevention and control of foodborne diseases. Methods Multiplicative seasonal ARIMA model was established based on the monthly foodborne disea- ses in Rudong from 2004 to 2010 ,and used foodbome diseases data of 2011 to verify the effect of model forecasting. The food- borne diseases in 2012 were predicted by ARIMA model based on the foodborne diseases from 2004 to 2010. Results Multi- plicative seasonal ARIMA （0,1,1） x （0,1,1） 12 better fit the trends of the number in previous time periods and series, with the mean of prediction fitting relative error of 2.7%. The number of foodborne disease in Rudong in 2012 was predicted to be 64. Conclusion Multiplicative seasonal ARIMA model can be used to fit the changes of foodborne diseases and to forecast the fu- ture foodborne disease. It is a predicted model of high precision for short time forecast.