Data Science & ML
Advanced

Bias-Variance Decomposition

Total expected error decomposes into bias, variance, and irreducible noise.

Formula

Error=Bias2+Variance+σ2Error = Bias^2 + Variance + \sigma^2

Variables

BiasUnderfitting error
VarianceOverfitting error
\sigma^2Irreducible

Example

High bias = underfit, high variance = overfit

Did You Know?

The bias-variance tradeoff guides model complexity choices.

theorygeneralization