广义回归神经网络在布鲁氏菌病预测中的应用
Application of generalized regression neural network in brucellosis prediction

目的:应用广义回归神经网络预测布鲁氏菌病月发病数,为布鲁氏菌病的防控提供科学依据。方法:通过Matlab 9.1软件创建广义回归神经网络,以2010年8月到2016年8月布鲁氏菌病的月发病人数为输入,2011年8月到2017年8月的月发病人数为输出,并预测2017年9至12月布鲁氏菌病的月发病人数。结果:广义回归神经网络的最优光滑因子取0.02。2017年9至12月布鲁氏菌病月发病人数真实值与预测值平均相对误差和决定系数分别为10.75%和0.64,模型拟合较好。结论:广义回归神经网络应用于布鲁氏菌病月发病人数的预测效果较好,可为该病的防控提供科学依据。
Aim:To forecast the monthly number of brucellosis applying the generalized regression neural network model,so as to provide scientific basis for the prevention and control of brucellosis.Methods:The generalized regression neural network was established by Matlab 9.1,the data of the monthly number of brucellosis in August 2010 to August 2016 was input, the monthly number from August 2011 to August 2017 was output, and data in September to December of 2017 were predicted.Results:The optimal spread of generalized regression neural network model was 0.02.The average relative error was 10.75% between the real and the predicted value from September to December of 2017, with R2 of 0.64,the trend was consistent,and the model was fitted well.Conclusion:Generalized regression neural network is effective in the prediction of brucellosis,which can provide scientific basis in prevention and control