ISTQB AI Testing ML Model Performance Metrics Quiz

1. Model bias in AI testing refers to systematic errors favoring certain groups. True or False?

True

False

2. Adversarial testing in AI involves evaluating model robustness against intentionally crafted inputs. True or False?

True

False

3. Which technique helps detect concept drift in deployed ML models?

Hyperparameter tuning

Continuous monitoring of performance metrics

Increasing training data size

Reducing model complexity

4. Model validation should only occur on the training dataset. True or False?

True

False

5. What does SHAP stand for in the context of model interpretability?

Shared Hypothesis Analysis Protocol

SHapley Additive exPlanations

Statistical Hierarchical Attribute Partitioning

Systematic Heuristic Assessment Process

6. Sensitivity and recall are synonymous terms in classification model evaluation. True or False?

True

False

7. Explainability in AI testing ensures models produce transparent, interpretable decisions. True or False?

True

False

8. Which metric is most appropriate for imbalanced datasets?

Accuracy

Precision and Recall

Mean Squared Error

R-Squared

9. A baseline model serves as a reference point for comparing more complex models. True or False?

True

False

10. What is the primary goal of fairness testing in AI systems?

Maximize accuracy regardless of group impact

Ensure equitable treatment across demographic groups

Reduce model complexity

Improve training speed

11. What metric measures the proportion of correct predictions out of all predictions made?

Accuracy

Precision

Recall

F1-Score

12. Which term describes when a model performs well on training data but poorly on new data?

Underfitting

Overfitting

Regularization

Normalization

13. Data drift occurs when input feature distributions change over time in production. True or False?

True

False

14. What does AUC (Area Under the Curve) represent in model evaluation?

Average Unit Cost

Probability of correct ranking of positive vs. negative cases

Absolute Uncertainty Coefficient

Algorithmic Uncertainty Calibration

15. Which cross-validation technique divides data into k equal-sized folds?

Stratified Cross-Validation

Leave-One-Out Cross-Validation

K-Fold Cross-Validation

Time-Series Cross-Validation

16. An ROC curve plots TPR against FPR to evaluate classifier performance. True or False?

True

False

17. What is the primary purpose of a confusion matrix in model evaluation?

To display training time

To show actual vs. predicted classifications

To measure data preprocessing quality

To identify overfitting automatically

18. The F1-Score is the harmonic mean of which two metrics?

Accuracy and Specificity

Precision and Recall

Sensitivity and Specificity

TPR and FPR

19. Recall measures the ability to identify all actual positive cases. True or False?

True

False

20. Which metric focuses on the proportion of true positives among all positive predictions?

Recall

Precision

Specificity

Accuracy

ISTQB AI Testing ML Model Performance Metrics Quiz

1. Model bias in AI testing refers to systematic errors favoring certain groups. True or False?

2.

What first name or nickname would you like us to use?

2. Adversarial testing in AI involves evaluating model robustness against intentionally crafted inputs. True or False?

3. Which technique helps detect concept drift in deployed ML models?

4. Model validation should only occur on the training dataset. True or False?

5. What does SHAP stand for in the context of model interpretability?

6. Sensitivity and recall are synonymous terms in classification model evaluation. True or False?

7. Explainability in AI testing ensures models produce transparent, interpretable decisions. True or False?

8. Which metric is most appropriate for imbalanced datasets?

9. A baseline model serves as a reference point for comparing more complex models. True or False?

10. What is the primary goal of fairness testing in AI systems?

11. What metric measures the proportion of correct predictions out of all predictions made?

12. Which term describes when a model performs well on training data but poorly on new data?

13. Data drift occurs when input feature distributions change over time in production. True or False?

14. What does AUC (Area Under the Curve) represent in model evaluation?

15. Which cross-validation technique divides data into k equal-sized folds?

16. An ROC curve plots TPR against FPR to evaluate classifier performance. True or False?

17. What is the primary purpose of a confusion matrix in model evaluation?

18. The F1-Score is the harmonic mean of which two metrics?

19. Recall measures the ability to identify all actual positive cases. True or False?

20. Which metric focuses on the proportion of true positives among all positive predictions?