Given a model that predicts (or estimates) a value, there are certain metrics that can be used for determining the quality of the model.

## Mean errors

These include:

- Mean Absolute Error, also called L1 Norm
- the simplest, calculates the average of the errors

- Mean Squared Error
- penalizes bigger errors, as the errors are squared

- Root Mean Squared Error
- has the same feature as MSE, but by squaring, it normalizes the value to be on the same scale as the y-vector

- Relative Absolute Error
- Relative Squared Error
- unaswered
- used to calculate $r_{2}$

These are calculated differently, but they are measuring the variance between actual values and predicted values, or in other words, they’re trying to find how big is the residual error of the model.