Problem with Model Deployment in Exel
  • Dear Gepsoft Team,

     

    I was trying to deploy esembled models to Excel, when I got the following

    Syntax error. Do you have any idea what could have caused the proplem?

    Many thanks for the answer in advance!

     

    I would have a question to esemble: Is it possible to do it only in Excel, or is their a way to acquire esembled source codes for C++ somehow from the program.

     

    Balazs

  • Hello,

    Thank you for reporting the bug. This happens when the operating system language uses commas as the decimal separator and the Rounding Threshold is zero. We are going to fix this very soon and we will be releasing an update for the application.

    For the moment, and as workaround, if you are using the User Defined rounding threshold then you can change it to a very small number (say 0,0000001), then select the menu History->Update All Thresholds and the deployment will work properly. If on the other hand you are using fitness function that force a rounding threshold of zero such as Maximum Likelihood, then I am afraid you will have to wait for the update.

    Another option is to temporarily change the Display Language in Regional Settings of the computer to English.
  • It is definitely possible to compose ensembles using the C++ code since you have access to the models translated to C++ in the Model Panel. You will have to compose the models yourself, though, following the example in the Excel deployment (calculate each model separately and then average the results).

    If you are creating small ensembles then it is doable but large ensembles with hundreds of models are impractical. This is in fact one of the features we are planning to add in a future release.
  • Yet another workaround for the zero rounding threshold problem when using the Maximum Likelihood fitness function is to temporarily change to another fitness function and then set the rounding threshold to a small number and then do the deployment since the fitness function is irrelevant at this stage.
  • Many thanks for the elaborate and fast reply! I do not need the solution right away, as I was trying only to experiment with the possibilities. It takes quite a while, until all the feature can be explored. It will be nice to have this feature.

    Many thanks again!

  • The bug has been fixed and a new version of GeneXproTools is ready for download in the download page.

    Again, thanks for reporting the bug and don't forget to uninstall GeneXproTools before installing this new version.

Howdy, Stranger!

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