Skip to content

ACCA Strategic Professional · Strategic Business Leader · Machine learning, AI and robotics

Brindle Bank trains a model on five years of labelled loan records, each marked 'defaulted' or 'repaid', so it can estimate default probability for new applicants. Which type of machine learning is being used, and what is the key implication for the bank?

It is supervised learning, because the model learns from historical records labelled with known outcomes. The implication is that accuracy and fairness depend on the quality and representativeness of that history, so biased data would lead to biased lending decisions.

  1. ASupervised learning; the quality and representativeness of the labelled history determines accuracy and any bias in decisionsCorrect
  2. BUnsupervised learning; the model will discover its own customer segments without historical outcomes
  3. CReinforcement learning; the model improves by receiving rewards from live trading outcomes
  4. DRule-based automation; the bank must code every lending criterion explicitly

Explanation

Training on records with known outcomes is supervised learning. Because the model learns from past decisions, unrepresentative or biased history will be reproduced in new decisions, creating governance and ethical risk. Unsupervised learning uses unlabelled data, so it does not fit.

Did you get it right without looking?

One question tells you little. A timed set on Machine learning, AI and robotics shows your real accuracy, how long you take and where you lose marks.

Practise Machine learning, AI and roboticsFree diagnostic mock

More Machine learning, AI and robotics questions