negative values
  • Dear Dr. Ferreira,
    when I deployment excel sheet
    How can I avoid negative values ​​in the output?

  • Hi,

    You can choose a fitness function that allows you to specify the output bounds, such as the Enhanced MSE fitness. In this case the learning algorithm will try to create models that fit into that range.

    For existing models though you can change the embedded Excel VBA code by hand, for example by adding the code below at the end of your gepModel function:

        If gepModel < 0.0 Then
            gepModel = 0.0           ' or some min value
        End If

    Another way is to 0/1 normalize your data as in this case, through the reverse normalization function in the Excel VBA code, the output is scaled back into the target range (it’s the training target range, so you might get out of range values for the test dataset or new scoring data).

    Candida Ferreira

  • Thank you so much for the elaborate and fast reply, Dr. Ferreira
  • Dear Dr. Ferreira,
    In some cases, I use 9  variables as input  and I notice that the program used only 7 variables in the model or less. How can I avoid that? Because I need all input variables in the model.  
  • Hi,

    You can evolve your models with Variable Pressure (it’s an adjustable parameter and you set it up in the Fitness Function Tab). Or you can just click the Complexify button in the Run Panel if you have a model that you think will benefit from integrating more variables.

    Candida Ferreira

  • Really, many thanks Dr. Ferreira for your efforts and fast reply.

Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!