Diagnosing Regularization Bias vs. Variance

Diagnosing Bias vs. Variance

This means:
  • High bias = under-fitting = Jtrain(θ) high and Jcv(θ) high
  • High variance = over-fitting = Jtrain(θ) low and Jcv(θ) high
Regularization: help prevent overfitting
This is the next step once we chose the degree of polynomial for our dataset

This means: 
  • Large λ = High bias = Jtrain(θ) high and Jcv(θ) high
  • Small λ = High variance = Jtrain(θ) low and Jcv(θ) high

Steps:


Resources:

Comments