Skip to content

FRM Part I · FRM Exam Part I · Machine-Learning Methods

A risk analyst fits a very flexible machine-learning model to predict loan defaults. The model achieves almost zero error on the training data but a much higher error on a held-out validation set. Which diagnosis is most appropriate?

The model is overfitting. Near-zero training error combined with much higher validation error shows it has memorized noise in the training data, giving low bias but high variance. Underfitting would instead produce poor performance on both the training and validation samples.

  1. AThe model is overfitting, with low bias and high varianceCorrect
  2. BThe model is underfitting, with high bias and low variance
  3. CThe model has high bias because it is too simple
  4. DThe validation set is too large, so the training error is unreliable

Explanation

A large gap between very low training error and much higher validation error indicates the model has fitted noise in the training sample. This is the signature of overfitting: low bias but high variance. Underfitting would show high error on both training and validation sets.

Did you get it right without looking?

One question tells you little. A timed set on Machine-Learning Methods shows your real accuracy, how long you take and where you lose marks.

Practise Machine-Learning MethodsFree diagnostic mock

More Machine-Learning Methods questions