A model achieves 99% training accuracy but only 72% validation accuracy. Which problem does this describe, and what is the most appropriate remedy?
Select an answer to reveal the explanation.
Short Explanation and Infographic
A model that aces training but bombs validation is like a student who memorized the practice test but can't answer any new questions — that's overfitting (high variance), and the fix is to constrain or get more data.
Full explanation below image
Full Explanation
The large gap between training accuracy (99%) and validation accuracy (72%) — a 27-point difference — is the hallmark of high variance / overfitting. The model has memorized the training data including noise, and fails to generalize. Remedies in order of preference: 1) Get more training data — the most reliable fix. 2) Regularization — L1 (Lasso) or L2 (Ridge) penalties, or dropout for neural networks. 3) Reduce model complexity — fewer trees, smaller max_depth, fewer hidden layers. 4) Feature selection — remove noisy features. 5) Ensemble averaging — bagging methods reduce variance. High bias (underfitting) would show both training AND validation accuracy being low and similar. Data leakage would actually inflate validation accuracy toward training accuracy, not reduce it.