error ??? Index exceeds matrix dimensions.

%

--------------------------------------------------------------------------
%define input and output for training
in=d(1:40,1:1000); 
ou=d(41:40,1:1000); 

trainInput= d(1:40,201:1000); % input for training
trainOutput= d(41:40,201:1000); % output for training

%--------------------------------------------------
% training the MLP Neural network 
% newff(input, target,[hidden layer1 hiddenlayer2 output],{transferFunction1 transferFunction2 transferFunction3});
% transferFunction defult for hidden layer is tansig but here use logsig
% and purelin is default transferFunction for output layer
net = newff(minmax(in),ou,[10 10 4],{'logsig' 'logsig' 'purelin'});           

%--------------------------------------------------------------------------
% Setting the parameters for training
net.trainParam.epochs = 1000; %set the maximum number of epochs to train
net.trainParam.goal = 0.02;  %sum-squared error goal.

%-------------------------------------------------------------------------- 
% training the MLP by using a a train function

net = train(net,trainInput,trainOutput); 

%--------------------------------------------------------------------------
% To draw the result making the testing 

testInput=d(1:40,1:200);   % extract certain column for testing input
testResult1=sim(net,testInput); %  simulate  the  MLP  network  by  using  a    sim function
testResult1 = testResult1';
trainOutput2= d(1:40,1:200);
??? Index exceeds matrix dimensions.

above is my code for fold 1 (train set is 201-1000, test set is 1-200) i use dataset 1000x40 double Any expert can u help me...thank you!!!

[kjetil87]

I didnt check the entire code but It migh be because of

ou=d(41:40,1:1000);
this is and empty matrix... Correct syntax is 

ou=d(41:-1:40,1:1000);

Answer 1

[Greg Heath]

Did it ever occur to you that it is self defeating to include code with a data set that is not accessible to those who want to help you?

In the future try your code on a MATLAB data set before posting. That way you can separate code errors from data errors and get a headstart on receiving focused help.

 help nndatasets
 doc nndatasets

Either use all of the chosen MATLAB dataset or a subset of it to best match the size of your data.

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

%define input and output for training

 in=d(1:40,1:1000); 
 ou=d(41:40,1:1000);

'1. ERRONEOUS INDEXING'

 trainInput= d(1:40,201:1000); % input for training
 trainOutput= d(41:40,201:1000); % output for training

'2.ERRONEOUS INDEXING'

%--------------------------------------------------

% training the MLP Neural network

 % newff(input, target,[hidden layer1 hiddenlayer2 output], ...
 % {transferFunction1 transferFunction2 transferFunction3});
 % transferFunction defult for hidden layer is tansig but here use logsig
 % and purelin is default transferFunction for output layer

 net = newff(minmax(in),ou,[10 10 4],{'logsig' 'logsig' 'purelin'});

'3. DOUBLY OBSOLETE FUNCTION. USE FITNET.'

'4. ONE HIDDEN LAYER IS SUFFICIENT'

'5. BIPOLAR TANSIG IS BETTER FOR HIDDEN LAYERS THAN UNIPOLAR LOGSIG'

'6. 4 OUTPUT NODES IS ERRONEOUS'

%--------------------------------------------------------------------------

 % Setting the parameters for training
 net.trainParam.epochs = 1000; %set the maximum number of epochs to train

'7. epochs = 1000 IS A DEFAULT. WHY BOTHER?'

 net.trainParam.goal = 0.02;  %sum-squared error goal.

'8. WHAT IS THE RATIONALE FOR THIS CHOICE??...(I ASSUME YOU MEANT MSE INSTEAD OF SSE)'

 %-------------------------------------------------------------------------- 
 % training the MLP by using a a train function
 net = train(net,trainInput,trainOutput); 
 %--------------------------------------------------------------------------
 % To draw the result making the testing 
 testInput=d(1:40,1:200);   % extract certain column for testing input
 testResult1=sim(net,testInput); %  simulate  the  MLP  network  by  using  a    sim function
 testResult1 = testResult1';
 trainOutput2= d(1:40,1:200);

'9.ERRONEOUS INDEXING'

THANK YOU FOR FORMALLY ACCEPTING MY ANSWER

GREG