Is there a way to change dividerand to divideblock in the Neural Time Series Toolbox?

I would like to use the time series neural net to predict future stock market closing prices using the NAR algorithm. I have loaded all the data values using the GUI, but the default setting for the GUI is to randomly divide the data points into learning, testing, and validation. I would like the first 50% to be dedicated to learning, the next 25% to validation, and the last 25% to testing against real values. Is there any way I could do that with the GUI? I am not too familiar with Matlab, so I would like to do this through the GUI, but if that is not possible then I will have to explore other options.

[Walter Roberson]

If I recall correctly, the GUI will generate code for you. You could generate the code and then edit it to use divideblock

[skahri1429]

I've done that but it references data that I have not defined in the command line. When I run it I get the error:

"Undefined function or variable 'Sheet 2_Table 1_1'.

Error in [function name] (line 9) T = tonndata(Sheet2_Table1_1, false, false);"

How would I import that into the function?

[Walter Roberson]

Could you attach the code it generates?

[skahri1429]

Attached

[Walter Roberson]

Change the variable name Sheet2_Table1_1 to the name of your time series data.

Then go down further and change the divide function and the ratios

Answer 1

[Greg Heath]

Using the command line:

net.divideFcn = 'divideblock';

Hope this helps.

Thank you for formally accepting my answer

Greg