MLA-C01 Practice Question: A data scientist has trained a model that…

A data scientist has trained a model that achieves 95% accuracy on the training set but only 70% on the test set. Which of the following is the most likely cause?

⚠ Common exam trap

The trap is that candidates confuse overfitting with data leakage — both can produce misleadingly high training metrics, but only overfitting shows the sharp train-high/test-low gap, while leakage typically inflates test performance too.

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

A large gap between high training accuracy (95%) and much lower test accuracy (70%) is the classic signature of overfitting: the model has memorized the training data, including its noise, and fails to generalize to unseen data. The model has learned patterns specific to the training set rather than the underlying distribution. Regularization, more data, or a simpler model would typically help.

Answer analysis

Option-by-option breakdown

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

✗
Data leakage
Why it's wrong here
Data leakage inflates training accuracy by exposing test or target information during training, but the stem shows a plain train-test gap with no contamination evidence, which indicates overfitting instead. Leakage is tempting because it also produces optimistic training scores, and it would be correct if validation data had entered the training pipeline.
✓
Overfitting
Why this is correct
Overfitting occurs when the model memorises training noise and patterns, producing high training accuracy but poor generalisation to unseen data. The 95% versus 70% gap directly reflects this memorisation, satisfying the stem's symptom of a large train-test performance disparity.
✗
Convergence to local minimum
Why it's wrong here
Convergence to a local minimum depresses training performance itself; it does not create a 25-point train-test gap. It is tempting because optimisation problems genuinely stall in local minima, and that would be the correct diagnosis if training loss had plateaued at an unsatisfactory level.
✗
Underfitting
Why it's wrong here
Underfitting produces poor accuracy on both training and test sets, so it cannot explain a 95% versus 70% gap. It is tempting because it is the opposite failure mode of overfitting, and would be the correct diagnosis if the model scored low on training data too.

Go deeper

MLA-C01 Practice Question: A data scientist has trained a model that…

Why each option matters

Option-by-option breakdown

Related to this question