Overfitting occurs when a model, insight, or conclusion is too closely aligned to a specific dataset and does not generalise to broader behaviour.
Overfitting reduces reliability and repeatability.
Examples in Action
- Optimising based on very small samples
- Over-segmenting audiences
- Drawing conclusions from isolated tests
- Excessive tuning without validation
Typical Outcomes / Results
- Reduced confidence in results
- Poor performance when scaled
- Misleading insights
- Increased risk of false positives
This definition reflects established analytical principles.