PEGACPDS88V1 Sample Questions & Answers
Revolves around adaptive model fundamentals and ongoing monitoring, the single biggest weight, plus building predictive models, prediction patterns within case management, introducing Pega AI concepts, NLP text analytics, process AI, and model governance.
Launch the full PEGACPDS88V1 simulator →Showing 10 of 20 free samples.
- Question 1Advanced
Adaptive Analytics · Monitoring and Impact Measurement
You are reviewing the 'Bubble Chart' in the Adaptive Model monitoring report. You observe a large bubble positioned in the top-right quadrant of the chart. What does this indicate about the predictor represented by this bubble?
Show answer & explanation
Correct answer: A
In the ADM Bubble Chart, the X-axis typically represents Performance (AUC) and the Y-axis (or bubble size/position context) represents Importance/Frequency. A bubble in the top-right indicates a 'Star' predictor: it is both highly predictive and applies to a large portion of the cases.
- Question 2Advanced
Adaptive Analytics · Adaptive Model Fundamentals
A Data Scientist is configuring an Adaptive Model and wants to ensure that two highly correlated predictors, 'AnnualIncome' and 'MonthlySalary', do not skew the model by double-counting the same signal. How does Pega's ADM automatically handle this situation?
Show answer & explanation
Correct answer: B
ADM performs automatic predictor grouping. It identifies correlated predictors and groups them. During scoring, it typically selects the 'best' predictor from each group to avoid multicollinearity issues and improve model robustness.
- Question 3Beginner
Adaptive Analytics · Monitoring and Impact Measurement
In the context of Adaptive Analytics, what is the significance of the 'Area Under the Curve' (AUC) metric value of 0.50?
Show answer & explanation
Correct answer: A
An AUC of 0.5 indicates a random classifier. It means the model cannot distinguish between positive and negative classes any better than flipping a coin. A good model typically has an AUC > 0.6 or 0.7.
- Question 4Intermediate
Adaptive Analytics · Adaptive Model Fundamentals
You need to configure an Adaptive Model to update its scoring logic. Which setting determines how frequently the Adaptive Decision Manager (ADM) refreshes the statistical models based on new evidence?
Show answer & explanation
Correct answer: C
While ADM collects data in real-time, the actual re-computation of the binning and probabilities (the 'update') happens periodically. This is controlled by system settings to balance performance and freshness, often defaulting to a set time interval or response count threshold.
- Question 5Advanced
Adaptive Analytics · Adaptive Model Fundamentals
A business requirement states that if a customer explicitly rejects an offer, this should be considered stronger negative evidence than if they simply ignored it. How can you configure the Adaptive Model to reflect this?
Show answer & explanation
Correct answer: A
In the Adaptive Model configuration, you define positive and negative behaviors. You can map 'Rejected' to the negative outcome category. While standard ADM treats all negatives similarly (binary target), advanced configurations or strategy logic can sometimes weight these, but primarily you ensure 'Rejected' is explicitly mapped as a negative outcome alongside 'Ignored'.
- Question 6Advanced
Adaptive Analytics · Adaptive Model Fundamentals
What happens to the existing binning of a numeric predictor in an Adaptive Model when new data arrives that falls significantly outside the current range of values?
Show answer & explanation
Correct answer: A
ADM uses dynamic binning. It automatically adapts the intervals (bins) for numeric predictors as new data flows in. If values fall outside the current range, the bin definitions are adjusted to include them, ensuring the model remains representative of the current data distribution.
- Question 7AdvancedSelect 2
Adaptive Analytics · Adaptive Model Fundamentals
Select TWO reasons why you would choose to enable 'Parameterize Predictors' in an Adaptive Model configuration. (Select TWO)
Show answer & explanation
Correct answers: C, D
Parameterized predictors allow you to pass context (like Region or Channel) as a parameter, enabling a single model definition to adapt to these contexts without creating thousands of separate model instances.
By parameterizing, you effectively create sub-segments within the model logic, allowing for more granular learning specific to that parameter combination (e.g., iPhone users in NY vs Android users in NY).
- Question 8Advanced
Adaptive Analytics · Monitoring and Impact Measurement
Case Study: A bank has been running an Adaptive Model for credit card offers for 6 months. The 'Income' predictor initially had a high Predictor Power (65%). However, over the last month, its power has dropped to 10%.
What is the most plausible explanation for this sudden drop?
Show answer & explanation
Correct answer: C
Adaptive models reflect current reality. If a predictor's power drops, it means it no longer splits the audience effectively. This could happen if the offer terms changed to be universally appealing, or if market conditions made income irrelevant for this specific decision.
- Question 9Beginner
Adaptive Analytics · Adaptive Model Fundamentals
True or False: An Adaptive Model in Pega MUST be trained with a historical dataset (CSV) before it can be deployed to production.
Show answer & explanation
Correct answer: B
False. One of the key benefits of Adaptive Models is that they do NOT require historical data. They can be deployed 'empty' and will learn from real-time interactions (Cold Start).
- Question 10Intermediate
Adaptive Analytics · Monitoring and Impact Measurement
You are analyzing the 'Response Count' vs 'Success Rate' in the Adaptive Model monitoring report. You see a model instance with 10,000 responses but a Success Rate of 0.01%. What business action should you consider?
Show answer & explanation
Correct answer: A
A high response count (lots of impressions/trials) with a near-zero success rate indicates the offer is irrelevant to the customers. Continuing to show it wastes valuable screen real estate. The model has 'learned' it's bad, but business logic should likely pull it.
Ready for the real thing?
The full PEGACPDS88V1 simulator has every exam-style question, timed mode, and instant scoring.