Question 19
Domain 2: ML Model DevelopmentA machine learning team compares two supervised models on the same dataset. Model A has low training error but much higher validation error. Model B has similarly high error on both training and validation sets. Which interpretation is most accurate?
Correct answer: A
Explanation
Overfitting is indicated when a model performs much better on training data than on validation data, while underfitting is indicated when performance is poor on both. The bias-variance tradeoff involves balancing model complexity so neither pattern dominates. — Bias-variance tradeoff, overfitting, underfitting detection.
Why each option is right or wrong
A. Model A is overfitting, and Model B is underfitting.
Model A shows low training error but much higher validation error, which matches overfitting. Model B shows similarly high error on both training and validation sets, which matches underfitting in this comparison.
B. Model A is underfitting, and Model B is overfitting.
Underfitting shows poor performance on both training and validation data, not low training error with high validation error.
C. Both Model A and Model B are overfitting.
Overfitting requires a gap where training performance is much better than validation performance.
D. Both Model A and Model B are underfitting.
Overfitting is identified by low training error paired with substantially worse validation error.