Description
If you take a machine learning course, at some point you will encounter the bias-variance trade-off. You probably feel like you sort of understand it, but not really. Later you see these two words…
Summary
- Bias and variance are often presented in technical terms, but they also have an intuitive interpretation.
- Such models are said to have high bias.
- So the sign of the error behaves like a biased coin.
- In practice step 2 is almost always skipped, because step 3 can already tell us if we are overfitting and to what extent, so step 2 acts more as a description of what would hypothetically happen if our model were overfitting.