Skip to main content
1-Visitor
January 7, 2022
Question

ThingWorx-Analytics: Creation of predictive model failed

  • January 7, 2022
  • 1 reply
  • 2035 views

Hello all,

 

I want to create a predictive model with an ordinal goal variable based on a time series dataset of around one million entries. Unfortunatly the creation always failed with the error you can find in the attachment. Does anybody know how I can fix this problem?

 

Here are some more information:

- ThingWOrx Version: ThingWorx 9.0.2

- Used learners in the model: Neural_Net, Decision_Tree and Gradient_Boost

- Used Ensambler: Average

- LookbackSize: 4

- No use of goal history

 

Thank you already in adcance.

 

Kind regards 

David

1 reply

19-Tanzanite
January 7, 2022

Hi David

 

 

From the screenshot you posted, the issue is due to an OutOfMemory error. This happens when working on large dataset.

Try the steps from https://www.ptc.com/en/support/article/CS294545 to increase the memory allocated to the workers

 

Thanks

Christophe

1-Visitor
January 18, 2022

Hi Christoph,

 

unfortunately I still get the same error after following the steps from the link you posted. Do you have any other idea how I can solve my issue?

Thank you for your support.

 

Best regards

David 

 

19-Tanzanite
January 18, 2022

Hi @DF_10027356 

 

Could you then download the TWALogCollector , execute it on the ThingWorx Analytics machine and post the resulting zip or gz file here ?

 

Thank you

Christophe