# Regression Metrics: MAE, RMSE and R² — Machine Learning Basics

Source: https://www.skillbyai.com/en/machine-learning/r-metrics

> Different metrics, different questions.

## Average error, big-error penalty, explained variance

**MAE** (mean absolute error) is the average size of errors in the target's units, easy to explain. **RMSE** (root mean squared error) also uses the target's units but punishes large errors more, useful when big misses are costly. **R²** compares the model with always predicting the mean: 1 is perfect, 0 is no better than the mean, negative is worse. Report a metric that matches the business cost of errors, in units people understand, and always next to a baseline.

## Delivery time estimates

An app that is usually 2 minutes off but sometimes 40 minutes off may have a decent MAE and a poor RMSE; customers remember the 40-minute misses.

## Explain errors in units

Say "on average 45 units off" rather than only quoting R2; stakeholders understand units.

**Quiz:** What does an R² of 0 mean?

- [ ] Every prediction is zero
- [ ] The model is perfect
- [x] The model is no better than always predicting the mean
- [ ] The data has no rows

*Answer:* The model is no better than always predicting the mean. R2 compares against the mean baseline.