What is Overfitting?
When a model learns its training data too closely and does badly on new data.
In plain English
An overfit model memorises noise and quirks in the training set. The warning sign is high accuracy on training data but much lower accuracy on data it has not seen. Remedies include more varied data, a simpler model, stopping training earlier and regularisation.
Example
A model scores 99% on its training set and 70% on a held-out test set.
Easy to confuse with
- Model drift: Overfitting shows up before release, when you compare training and test results. Drift appears after release as the world changes.
Where you will meet it
This idea belongs to the subject area of these practice tests. We do not publish or predict real exam questions.
- AWS MLA-C02 practice tests
- NVIDIA NCA-AIIO practice tests
- PMI-CPMAI practice tests
- AWS AIF-C01 practice tests
Related terms
- Supervised vs unsupervised learning: Supervised learning trains on labelled examples. Unsupervised learning looks for structure in unlabelled data.
- Model drift: A model getting worse over time because the real world no longer looks like its training data.
- Fine-tuning: Training an existing model a little more on your own examples so it behaves in the way you want.
Test yourself
Practice tests with an explanation for every answer option, so each mistake teaches you something.
Browse practice testsAll glossary termsWritten and checked by the Be Ready Education team, last reviewed 2 November 2026. General study information, not professional advice. Providers change names and rules, so confirm details on the official documentation. Not affiliated with, endorsed by or sponsored by any certification vendor or exam body. Certification names, exam codes and product names are trademarks of their respective owners and are used only to identify the exam a practice product prepares for. Our questions are original and are not actual exam questions.