Question
Best practice while building models in PAD
Hi,
I am looking for an advice on the best practice while building models in Predictive Analytics Director:
Would it be reasonable to do it directly in PROD environment or it is suggested to run this action in a different environment in order not to impact the PROD performance?
Can you please also specify, which nodes/resources are impacted during this stage? PAD nodes for sure, but any of the below as well?
- DDS nodes ?
- DB?
- Data flow nodes?
Thank you in advance for your help.
***Moderator Edit-Vidyaranjan: Updated Platform Capability***
Hi Katarzyna,
Building offline models using PAD is usually not done on a PROD environment. It uses Pega nodes and as far as I know doesn't rely on specific decisioning nodes (unless training data is sourced from a data flow).
BTW building predictive models using Pega machine learning is now done directly from Prediction Studio,introduced in 8.1, we tend not to refer to 'PAD' as a separate module anymore because this functionality is now embedded,
Regards,
Ivar