# Bias-Variance Tradeoff

*noun*

say it: **“BY-us VAIR-ee-unss TRAYD-awf”** **/ˈbaɪəs ˈvɛriəns ˈtreɪdˌɔf/**

The balance in machine learning between a model too simple to capture a pattern and one so complex that it also captures random noise.

## What is the **bias-variance tradeoff**?

The **bias-variance tradeoff** is the balance between a model too simple to catch the pattern (high bias:
underfitting) and one so flexible it learns each example's noise (high variance: overfitting). Making a model more
flexible lowers one and raises the other, so the best setting sits in between.

## In a sentence

- “The validation curve made the **bias-variance tradeoff** easy to see.”
- “Picking the tree's depth is the **bias-variance tradeoff** in action: too shallow misses the pattern, too deep memorizes noise.”

Level: Freelancer · [More AI Basics terms at this level](https://www.ask-codey.com/glossary?category=ai&level=4)

Updated: November 19, 2026

Source: https://www.ask-codey.com/glossary/bias-variance-tradeoff
