知识与财富的链接
目的 比较自回归移动平均(ARIMA)模型与广义回归神经网络(GRNN)模型对于肺结核发病率的预测性能.方法 根据我国2004年1月至2012年12月的肺结核逐月发病率数据资料,应用Eviews 7.0.0.1建立ARIMA模型,应用Matlab 7.1的神经网络工具箱建立GRNN模型;选取2013年肺结核逐月发病率数据对两种预测模型进行检验,比较预测结果.结果 ARIMA模型和GRNN模型的Theil不等系数(TIC)分别是0.034和0.059,说明ARIMA模型对我国2013年肺结核逐月发病率的拟合程度优于GRNN模型,ARIMA模型相对误差绝对值仅为GRNN模型的57.19％.结论 ARIMA预测模型更适合用于我国肺结核发病率的预测;建议尝试组合模型预测肺结核发病率.
Objective : To compare the performance of ARIMA model and GRNN model for predicting the incidence of tuberculosis. Methods: Set up ARIMA model by Eviews7.0.0.1 and GRNN model by neural network toolbox of Matlab7.1 according to monthly tuberculosis incidence data from January 2004 to December 2012 in China. Compare prediction of 2013 monthly tuberculosis incidence data with both models. Results: The TIC are 0.034 and 0.059 for ARIMA model and GRNN model respectively, indicating that ARIMA model is better than GRNN model to fit with 2013 the monthly incidence of tuberculosis. The absolute value of the relative error for ARIMA model is only 57.19% of GRNN model. Conclusion : ARIMA prediction model is more suitable for the incidence of TB. We suggest to try a combination of models predict the incidence of infectious diseases.
