Skip to content

IAI Actuarial Core Principles · Risk Modelling and Survival Analysis · Elementary principles of machine learning

A pricing team fits a very flexible model to claim data and finds that its error on the training data is very small but its error on a separate test set is much larger. Which statement best describes this situation?

The model is overfitted. Very low training error combined with much higher test error shows the model has memorised noise in the training data and does not generalise, whereas underfitting would produce poor performance on both the training and test sets.

  1. AThe model is overfitted, as it has captured noise specific to the training dataCorrect
  2. BThe model is underfitted, as it is too simple to capture the pattern in the data
  3. CThe model has high bias and low variance
  4. DThe test set is too large and should be merged into the training data
  5. The model is well calibrated, as training error is the best measure of future performance

Explanation

A large gap between low training error and high test error is the classic sign of overfitting: the model has learnt noise that does not generalise. Underfitting would show high error on both sets. High bias with low variance describes a too-simple model, not this case.

Did you get it right without looking?

One question tells you little. A timed set on Elementary principles of machine learning shows your real accuracy, how long you take and where you lose marks.

Practise Elementary principles of machine learningFree diagnostic mock

More Elementary principles of machine learning questions