Intermediate · Safety
Coverage bias
Visual diagram · (in preparation) · Math · (in preparation) · Worked example · 3 difficulty levels.
TL;DR. Coverage bias occurs when a model is trained on data that does not accurately represent the full population it will be used on.
Technical Definition
Coverage bias occurs when a model is trained on data that does not accurately represent the full population it will be used on.
How it works
Coverage bias is a type of selection bias where the training data does not adequately cover the diversity of the real-world population the model will encounter. This can lead to poor performance or unfair outcomes for underrepresented groups. It highlights the importance of representative datasets in machine learning.
Related Concepts
- Bias (ethics/fairness) — Unfair prejudice or favoritism towards certain groups or things, which can influence data, system design, and user interactions.
- Data set or dataset — A dataset is a collection of raw data, often organized in formats like spreadsheets or CSV files.