论文部分内容阅读
目的探讨ARIMA-NARNN组合模型预测血吸虫感染率的有效性。方法利用2005年1月至2015年2月江苏省血吸虫感染率资料分别建立ARIMA模型、NARNN模型和ARIMA-NARNN组合模型,比较各模型的拟合和预测效果。结果相比较ARIMA模型和NARNN模型,ARIMA-NARNN组合模型预测样本的MSE、MAE和MAPE均最小,分别为0.011 1、0.090 0和0.282 4。结论 ARIMA-NARNN组合模型能有效模拟和预测血吸虫感染率,具有较好的推广应用价值。
Objective To investigate the effectiveness of ARIMA-NARNN combination model in predicting schistosoma infection. Methods The ARIMA model, NARNN model and ARIMA-NARNN combination model were established respectively from January 2005 to February 2015 in Jiangsu Province. Fitting and prediction results of each model were compared. Results Compared with the ARIMA model and the NARNN model, the ARIMA-NARNN combined model predicts the smallest MSE, MAE and MAPE, which are 0.011, 1.0900 and 0.282 4, respectively. Conclusion ARIMA-NARNN combination model can effectively simulate and predict the infection rate of schistosomiasis, which has a good promotion and application value.