Regression Metrics

← Back to Model Evaluation

Metrics for evaluating models that predict continuous values. Each metric has different sensitivity to outliers and different interpretability.

Key Metrics

• MSE (Mean Squared Error) — average of squared differences, penalizes large errors
• RMSE (Root MSE) — same units as target, more interpretable
• MAE (Mean Absolute Error) — average of absolute differences, robust to outliers
• R-squared — proportion of variance explained
• Adjusted R-squared — R-squared penalized for number of features

Related

• Classification Metrics (discrete outputs)
• Linear Models (commonly evaluated with regression metrics)

---

ml evaluation regression metrics