ANFIS: Out-of-sample forecasting

Hello,

Please help me with the following:

Consider a TrainInput matrix

A=

[6445.63 6341.97

6554.53 6445.63

6697.14 6554.53

6867.47 6697.14 ];

Consider a TrainOutput matrix

B=

[6554.53

6697.14

6867.47

6965.699

6986.04];

During training, the TrainOutput becomes TrainInput when moving tho the next sample.

For instance,

TrainOutput(1)= 6554.53;

becomes

TrainInput(2,1) in order to forecast TrainOutput(2).

When reaching at the end of TrainInput matrix, I need to forecast the next 10 sample using as training samples the ANFIS outputs (there is no available TestInput).

I`ve tried using the following code:

fis = anfis([TrainInputs,TrainTargets]);

TrainOutputs=evalfis(TrainInputs,fis);

for i=1:10

TrainOutputs(4+i)=evalfis([TrainInputs(4+i);[TrainOutputs(end),TrainInputs(end,1)]],fis);

end

but it does not work.

Thank you.

Best,

Pavlos