Certified Pega Data Scientist (PEGACPDS24V1) Free Sample Questions

In Pega Customer Decision Hub, the standard Next-Best-Action arbitration formula prioritizes actions by multiplying Propensity (likelihood to accept), Context weighting (relevance in the moment), Action value (financial impact), and Business levers (strategic priority). This comprehensive approach ensures that both customer needs and business objectives are balanced in real-time.

Predict Churn Risk (or Likelihood to Churn) is a standard prediction used to assess how likely a customer is to leave. This propensity is then used in arbitration to prioritize retention offers over cross-sell offers for at-risk customers.

False. Engagement policies are evaluated first to filter the set of actions. AI models (propensity calculations) and arbitration are only applied to the subset of actions that pass the Eligibility, Applicability, and Suitability rules. This saves computational resources and ensures business compliance.

Pega's Adaptive Decision Manager (ADM) fundamentally relies on a Naive Bayes algorithm and Bayesian scoring techniques. This approach is highly efficient for online learning, allowing the model to instantly update predictor bins and probabilities as each new customer response (accept/reject) is recorded.

A high success rate combined with a low AUC (near 50) means that almost every customer is accepting the offer, making it impossible for the model to find differentiating predictors. This often happens when an offer is 'too good to be true' (e.g., free money), indicating the business might be giving away value unnecessarily.

Predictor grouping in ADM identifies highly correlated predictors (e.g., 'Age' and 'Date of Birth') and groups them together. When calculating propensity, the model only uses the single most predictive active variable from each group. This prevents the Naive Bayes algorithm from double-counting correlated evidence, which would skew the propensity score.