If a model has a large deviation on the test set and a small variance, it means that the model?

Question

Options

• AOverfitting
• BMay be overfitting may be underfitting
• CJust fit
• DUnderfitting

• A
  13% (7)
• B
  4% (2)
• C
  5% (3)
• D
  79% (44)

Explanation

Large deviation on the test set means the model's predictions are far from the true values (high bias), while small variance means the model produces consistent but consistently wrong predictions regardless of the input data. This combination - high bias, low variance - is the textbook definition of underfitting, where the model is too simple to capture the underlying pattern in the data.

Option A (overfitting) is wrong because overfitting produces low bias on training data but high variance - the model memorizes noise and performs inconsistently on unseen data. Option B is wrong because the specific pairing of high deviation and low variance is not ambiguous; it points unambiguously to underfitting. Option C (just fit) is wrong because a well-fitted model would have both low deviation and low variance.

Memory tip: Pair the two terms with their problems - high Bias means the model is far off (think B for "Bad aim, every shot misses the same way"), high Variance means it is unstable. Large deviation maps to high bias, which always signals underfitting.

No community discussion yet for this question.