Skip to content

ACCA Strategic Professional · Advanced Performance Management · Technology and information systems

Orion Logistics trained a machine learning model on 24 months of delivery data. It achieves very high accuracy on the training data but performs poorly when predicting costs for new routes. Which problem is most likely, and what is the appropriate remedy?

The most likely problem is overfitting: the model has learned noise specific to the training data and so fails on new routes. The remedy is to validate on unseen data and simplify the model or supply more varied data. Underfitting would show weak results on the training data too.

  1. AOverfitting; validate on unseen data and simplify the model or add more varied dataCorrect
  2. BUnderfitting; add more complex features and train on the same data only
  3. CData lake failure; move the data into a data warehouse
  4. DBlockchain latency; increase the number of network nodes

Explanation

Excellent fit on training data but poor results on new data is the classic sign of overfitting, where the model captures noise rather than general patterns. Remedies include validating on a separate test set, simplifying the model and using more varied data. Underfitting would show poor performance on both training and new data.

Did you get it right without looking?

One question tells you little. A timed set on Technology and information systems shows your real accuracy, how long you take and where you lose marks.

Practise Technology and information systemsFree diagnostic mock

More Technology and information systems questions