grnn神经网络的数据怎么选择

GRNN的网络结构

代码例子： % GRNN的数据预测—基于广义回归神经网络的货运量预测 %% 清空环境变量clc;clear allclose allnntwarn off; %% 载入数据load data;% 载入数据并将数据分成训练和预测两类p_train=p(1:12,:);t_train=t(1:12,:);p_test=p(13,:);t_test=t(13,:);%% 交叉验证desired_spread=[];mse_max=10e20;desired_input=[];desired_output=[];result_perfp=[];indices = crossvalind('Kfold',length(p_train),4);h=waitbar(0,'正在寻找最优化参数....');k=1;for i = 1:4    perfp=[];    disp(['以下为第',num2str(i),'次交叉验证结果'])    test = (indices == i); train = ~test;    p_cv_train=p_train(train,:);    t_cv_train=t_train(train,:);    p_cv_test=p_train(test,:);    t_cv_test=t_train(test,:);    p_cv_train=p_cv_train';    t_cv_train=t_cv_train';    p_cv_test= p_cv_test';    t_cv_test= t_cv_test';    [p_cv_train,minp,maxp,t_cv_train,mint,maxt]=premnmx(p_cv_train,t_cv_train);    p_cv_test=tramnmx(p_cv_test,minp,maxp);    for spread=0.1:0.1:2;        net=newgrnn(p_cv_train,t_cv_train,spread);        waitbar(k/80,h);        disp(['当前spread值为', num2str(spread)]);        test_Out=sim(net,p_cv_test);        test_Out=postmnmx(test_Out,mint,maxt);        error=t_cv_test-test_Out;        disp(['当前网络的mse为',num2str(mse(error))])        perfp=[perfp mse(error)];        if mse(error)