伤寒副伤寒发病趋势的ARIMA模型和GM（1,1）模型预测

目的采用求和自回归滑动平均模型（ARIMA模型）和灰色模型GM（1,1）分别对常州市伤寒副伤寒发病率进行预测,并比较两者预测的准确性。方法根据2001—2010年的月发病率数据分别建立ARIMA模型和GM（1,1）模型。通过比较2011年实际发病率和模型预测发病率间的相对误差验证模型预测的准确性。采用准确性高的模型预测2012—2013年伤寒副伤寒的年发病率。结果 GM（1,1）模型对伤寒副伤寒的年发病率预测准确率较高。2012年和2013年伤寒副伤寒预测的年发病率分别为2.52/105和2.33/105。结论对于伤寒副伤寒发病率的预测,应同时拟合几种模型,选择拟合效果最好的一种模型。

A Prediction Study for the Incidence Trend of Typhoid and Paratyphoid Fever with ARIMA Model and GM (1, 1) Model

Objective To predict the typhoid and paratyphoid incidence trend in Changzhou city based on the ARIMA model and grey system model GM （ 1, 1 ）, and to compare the accuracy of the two models. Methods ARIMA model and GM （ 1, 1 ） model were set up respectively based on the data on the month incidence during the years of 2001--2010. By comparing the relative error between the actual incidence and the predictive incidence from the models, the accuracy of the prediction based on the models was verified. The model with higher accuracy was used to predict the typhoid and paratyphoid incidence during 2012--2013. Results GM （ 1, 1 ） model was more accurate on predicting the typhoid and paratyphoid incidence. The predictive typhoid and paratyphoid incidence was 2. 52/105 in the 2012 and 2. 33/10^5 in the 2013. Conclusion When predicting the typhoid and paratyphoid incidence, several model should be fitted to choose the model with the best fitting effect.