Question
Areteans Technology Solutions Australia Pty Ltd.
IN
Last activity: 21 Jun 2023 8:02 EDT
Multiclass Classification Problem in Process AI
Can we solve a ML Problem for Pega Process AI where Target is a "Multiclass Clasification" type (e.g predict to highest selling product(P1,P2,P3,P4 etc) from a set of products) based on some list of prdictors ?
This query arisis as while creating/importing a Model in Pega , it allows to select target variable type as "continuous" or "binary classification" (like Fraud determination)?
@Saleem_A @Otto_Perdeck @Peter van der Putten - Can you kindly provide any suggestions ?
Thanking you in advance.
Thanks
Chandan
Accepted Solution
Updated: 21 Jun 2023 8:02 EDT
Pegasystems Inc.
IN
@MarijeSchillern Currently process AI supports binary value (Yes/No, Churn/ No churn/ Click/ No Click) predictions using adaptive modeling and predictive modeling, and also supports continuous value prediction using predictive modeling. But the use case @CHANDANPAN is mentioning is of multi-class prediction or categorical modeling use case which is not supported in process AI yet.
We can use predictive models though for the same using a PMML or H2O or any cloud-based ML engine models like Amazon sagemaker or Google
Accepted Solution
Updated: 21 Jun 2023 8:02 EDT
Pegasystems Inc.
IN
@MarijeSchillern Currently process AI supports binary value (Yes/No, Churn/ No churn/ Click/ No Click) predictions using adaptive modeling and predictive modeling, and also supports continuous value prediction using predictive modeling. But the use case @CHANDANPAN is mentioning is of multi-class prediction or categorical modeling use case which is not supported in process AI yet.
We can use predictive models though for the same using a PMML or H2O or any cloud-based ML engine models like Amazon sagemaker or Google