论文部分内容阅读
目的:比较Elman神经网络模型与自回归移动平均(ARIMA)模型对流感发病率预测的效果。方法:选取河南省2006年1月至2010年12月的流感疫情资料作为训练集,2011年1月至12月的流感疫情资料作为检验集,前者用于Elman神经网络模型和最优ARIMA模型的建立,后者用于两种模型的预测效能的检验与评价。结果:在最优ARIMA(1,0,0)模型和最优Elman神经网络预测模型下,检验集预测值的平均误差绝对值、平均误差绝对率和非线性相关系数分别为0.133、0.238、0.708和0.152、0.271、0.725。结论:Elman神经网络模型具有与ARIMA模型相近的预测效能,在流感发病率预测中有较好的应用价值。
Objective: To compare the effect of Elman neural network model and autoregressive moving average (ARIMA) model on the prediction of influenza morbidity. Methods: Influenza epidemic data from January 2006 to December 2010 in Henan Province were selected as the training set. Influenza epidemic data from January to December in 2011 were selected as the test set. The former was used in Elman neural network model and the best ARIMA model The latter is used to test and evaluate the predictive performance of the two models. Results: Under the optimal ARIMA (1, 0, 0) model and the optimal Elman neural network prediction model, the absolute value of the average error of the test set predictive value, the mean absolute error rate and the non-linear correlation coefficient were 0.133,0.238,0.708 And 0.152, 0.271, 0.725. Conclusion: The Elman neural network model has similar predictive performance as the ARIMA model, and has good application value in the prediction of flu morbidity.