How to export predicted multi-output data from trained neural network?

Hi, I have trained a feedforward neural network and would like to use it to simulate and predict the output using new input data. However, my code only able to predict single output data. Any thing wrong with my coding for simulation of neural network (code that highlight in BOLD)? How to export multioutput datas to workspace by using coding? Below is my code for neural network. Can anybody help me to solve this?

Develop Neural Network Model

%Import data

inputs=dlmread('Inputs2.txt', '\t', 1, 0); %input data

targets=dlmread('Targets2.txt', '\t', 1, 0); %target data

inputrep=repmat(inputs,4,1); %replicate input data

targetrep=repmat(targets,4,1); %replicate target data

inputrep = inputrep'; %transpose the data

targetrep = targetrep';

[inputrepn, inputreps]=mapminmax(inputrep); %normalize the data

[targetrepn, targetreps]=mapminmax(targetrep);

trainFcn = 'trainlm'; %use the Levenberg-Marquardt algorithm

%Train the networks

for i=1:30 %vary number of hidden layer neurons from 1 to 100

hiddenLayerSize = i; %number of hidden layer neurons

net = fitnet(hiddenLayerSize,trainFcn); %create a fitting network

net.divideParam.trainRatio = 70/100; %use 70% of data for training

net.divideParam.valRatio = 15/100; %15% for validation

net.divideParam.testRatio = 15/100; %15% for testing

[net,tr] = train(net,inputrepn,targetrepn); % train the network

outputs = net(inputrepn(:,tr.testInd)); %simulate 15% test data

rmse15(i)=sqrt(mean((outputs-targetrepn(tr.testInd)).^2)); %RMSE for 15% random test data

r15(i)=regression(targetrepn(tr.testInd), outputs);

save(['networks\net' num2str(i)],'net'); %save the network in networks folder

end

%Plot the RMSEs

plot(1:30, rmse15, 'b*-');

legend('Random'); xlabel('Number of hidden layer neurons');

ylabel('RMSE (^oC)');

%Save the RMSEs

fid=fopen('rmse.txt', 'wt');

fprintf(fid, 'Nh\t RMSE15\t');

fprintf(fid, '%4.0f\t %f\t %f\n', [1:30; rmse15]);

fclose all;

Simulate Neural Network Model

%Import data

inputnew =dlmread('Inputs3.txt', '\t', 1, 0); %input data

inputnew = inputnew'; %transpose the data

[inputnewn, inputnews]= mapminmax(inputnew); %normalize the data

%simulate network to predict output

yn = sim(net,inputnewn);

y = mapminmax('reverse',yn,targetreps); %denormalize data

y = y'; %transpose the data

%Save the predicted outputs

fid=fopen('y.txt', 'wt');

fprintf(fid, 'Yield\t O/C\t C/N\t');

fprintf(fid, '%4.0f\t %4.0f\t %4.0f\t %f\n', [1:14; rmse15]);

fclose all;

Answer 1

[Mahesh Taparia]

Hi

To predict the model on test data, use 'predict' function instead of 'sim'. For more information, refer to the documentation of predict function.