基于ARIMA的猪丹毒预测模型研究

Estabishment of an ARIMA model for prediction of swine erysipelas

为研究ARIMA时间序列模型预测猪丹毒(SE)的可行性,本实验选取兽医公报发布的我国某省SE的发病数据,对2005年1月～2009年6月该地区的SE发病资料进行模型构建与拟合,采用构建的模型对2009年7月～2009年12月的SE发病率进行预测并验证预测效果。结果显示ARIMA(2,1,0)模型可以拟合既往时间段上的发病率序列,预测结果显示MSE为0.203×10-10、MAPE为0.293,该模型可以预测未来SE的发生。

The objective of this study is to clarify the feasibility of using ARIMA model to predict the incidence of swine erysipelas （SE）. An ARIMA model was established based on the incidence data of SE in the areas from Jan. 2005 to Jun. 2009 according to the Veterinary Bulletin. Then the constructed model was used to predict the monthly incidence from Jul. 2009 to Dec. 2009 and the prediction was validated by comparing with the actual incidences. ARIMA（2, 1, 0） exactly fitted the incidence of the previous monthly incidence from Jan. 2005 to Jun. 2009, the predicted monthly incidence in Jul. 2009 to Dec. 2009 by the model was consistent with the actual incidence and the MSE was 0.203 × 10-10, the MAPE was 0.293. The model could be use to fit exactly the changes of SE incidence and to predict the incidence trends in future.