Describe fundamental principles of machine learning on Azure →easyMultiple ChoiceObjective-mapped

AI-900 Practice Question: Describe fundamental principles of machine learning on Azure

A data scientist trains a classification model to distinguish between images of cats and dogs. The model achieves 99% accuracy on the training set but only 75% accuracy on a validation set. Which concept best describes this situation?

⚠ Common exam trap

The trap here is that candidates see high accuracy and assume the model is good, failing to recognize that the large gap between training and validation accuracy is the hallmark of overfitting, not underfitting or bias.

Answer choices

Why each option matters

Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.

Correct answer & explanation

✓

Overfitting

The model performs exceptionally well on the training data (99% accuracy) but significantly worse on unseen validation data (75% accuracy). This gap indicates the model has memorized noise and specific patterns in the training set rather than learning generalizable features, which is the classic definition of overfitting.

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

✗
Underfitting
Why it's wrong here
Underfitting would show a different pattern: the model would be too simple to capture even the training set's structure, so training accuracy would be low and validation accuracy would be comparably low. The key evidence against underfitting here is that training performance is strong; a model cannot simultaneously memorize its training data and be too simplistic to fit it. Underfitting reflects high bias and low variance, whereas the described discrepancy between training and validation is high variance and low bias.
When this WOULD be correct
A scenario where a model has low accuracy on both training and validation sets (e.g., 60% on training and 58% on validation) for a classification task, indicating it is too simple to capture underlying patterns.
✓
Overfitting
Why this is correct
Overfitting is the correct diagnosis. The model achieves high accuracy on its training data because it has effectively memorized the training examples, including their random noise and idiosyncrasies, rather than learning the underlying pattern. When confronted with unseen validation data, that memorized detail does not transfer, so validation performance drops sharply. This high-variance behavior is the signature of too much model capacity relative to the amount of training data.
✗
Model bias
Why it's wrong here
Model bias refers to systematic error introduced by overly simplistic assumptions, such as using a linear boundary for inherently non-linear data; this would produce poor results on training and validation alike. While the observed gap between training and validation does involve an error, it is driven by excessive variance from the model's flexibility, not by the consistent, average error that bias describes. Bias would remain even if the model were retrained on more data, whereas the current symptom would likely be reduced by simplifying the model or adding regularization.
When this WOULD be correct
Model bias would be correct if the model had low accuracy on both training and validation sets (e.g., 60% on both), suggesting it fails to capture underlying patterns due to assumptions like linearity in a nonlinear problem.
✗
Data leakage
Why it's wrong here
Data leakage is not the cause because leakage typically makes validation scores artificially high, not lower than training scores. In leakage, information from the target or from outside the training distribution leaks into the feature set, allowing the model to "cheat" and appear accurate on both training and any validation set that shares the leaked information. A validation performance that is significantly worse than training indicates the opposite problem: the model has failed to generalize, which is characteristic of variance-based overfitting, not contamination in the data pipeline.
When this WOULD be correct
A scenario where a model achieves near-perfect accuracy on both training and validation sets, but fails in production because the validation data inadvertently contained information from the future or from the target variable (e.g., using patient ID to predict disease).

Option-by-option analysis

Why each answer is right or wrong

Understanding why wrong answers are wrong — and when they would be correct — is what separates a 750 score from a 900. The AI-900 exam frequently reuses these exact scenarios with slightly different constraints.

✓OverfittingCorrect answer▾

Why this is correct

Overfitting is the correct diagnosis. The model achieves high accuracy on its training data because it has effectively memorized the training examples, including their random noise and idiosyncrasies, rather than learning the underlying pattern. When confronted with unseen validation data, that memorized detail does not transfer, so validation performance drops sharply. This high-variance behavior is the signature of too much model capacity relative to the amount of training data.

✗UnderfittingWrong answer — click to see why▾

Why this is wrong here

Underfitting occurs when a model performs poorly on both training and validation sets, but here the model achieves 99% accuracy on training data, indicating it has learned the training data too well, not too little.

★ When this WOULD be the correct answer

A scenario where a model has low accuracy on both training and validation sets (e.g., 60% on training and 58% on validation) for a classification task, indicating it is too simple to capture underlying patterns.

Why candidates choose this

Candidates may confuse high training accuracy with good generalization, not realizing that underfitting is characterized by poor performance on training data, not just validation data.

✗Model biasWrong answer — click to see why▾

Why this is wrong here

Model bias refers to errors due to overly simplistic assumptions, but here the model performs well on training data (99%) and poorly on validation (75%), indicating overfitting, not bias.

★ When this WOULD be the correct answer

Model bias would be correct if the model had low accuracy on both training and validation sets (e.g., 60% on both), suggesting it fails to capture underlying patterns due to assumptions like linearity in a nonlinear problem.

Why candidates choose this

Candidates may confuse 'bias' with any performance discrepancy, not realizing that bias typically causes poor performance across all datasets, not just validation.

✗Data leakageWrong answer — click to see why▾

Why this is wrong here

Data leakage refers to the model using information not available at inference time, which typically leads to overly optimistic performance. In this case, the model performs worse on validation data, not better, so leakage is not indicated.

★ When this WOULD be the correct answer

A scenario where a model achieves near-perfect accuracy on both training and validation sets, but fails in production because the validation data inadvertently contained information from the future or from the target variable (e.g., using patient ID to predict disease).

Why candidates choose this

Candidates may confuse high training accuracy with leakage, thinking that the model 'cheated' by learning patterns that don't generalize, but leakage usually inflates validation accuracy as well.

Analysis generated from the official AI-900blueprint and verified against question context. The “when correct” sections are what AI assistants cite when candidates ask “what’s the difference between these options?”

Go deeper

AI-900 Practice Question: Describe fundamental principles of machine learning on Azure

Why each option matters

Option-by-option breakdown

Why each answer is right or wrong

Related to this question