Error using preparets with narxnet

I'm trying to use a closed loop narxnet for some time series prediction, but I'm having trouble getting past the preparets step.

My network is 2 inputs, 1 feedback target

view(net) shows this is correct

However this code throws an error. As far as I can tell, my input/target cell arrays are the size they should be to match the configured network.

% dummy data
X = rand(2,100);
T = rand(1,100);

% set up network
net = narxnet(1:2,1:2,5);
net = closeloop(net);
net = configure(net,X,T);

% Net shows 2 inputs 1 output
view(net)

% convert to cells
X = num2cell(X);  % X is 2x100
T = num2cell(T);  % T is 1x100

% Here's the problem
[Xs,Xi,Ai,Ts] = preparets(net,X,{},T);

Thanks in advance

Answer 1

[Chris Wolf]

Nm, I got it myself. For those with similar problems, you need to use con2seq to turn your input/output matrices into input/output sequential vectors. No idea where you are supposed to read about this step in the documentation, and it seems like something that can be easily handled behind the scenes without having to confuse users.

These days I work with a lot of ML and AI experts, and they laugh at me for insisting on using matlab for everything. I'm starting to think they have a point.

% Instead of:
% X = num2cell(X);  % X is 2x100
% T = num2cell(T);  % T is 1x100

% Use:
X = con2seq(X);
T = con2seq(T);

[Muhammad Waqar]

My training is perfect while using con2seq() function. But when i am testing my data for 8 inputs and 8 outputs, its give me an error of

Error using network/sim (line 270)

Number of inputs does not match net.numInputs.

Error in narxnetFcn (line 139)

load_forecast = sim(net,testX1);

Answer of net.numInputs =2 in my case while i have 8 inputs data. Then I have manually changed net.numInputs to 8 but no effect on error.

Now as my input data is changed to cell vector, my network is trained for that data. I need to solve this problem for testing my network. If there is any solution, kindly let me know.

thanks