Build the mental model
MAE, RMSE and R² summarise different aspects of prediction error. Residual plots reveal structure that a single scalar cannot. Report a metric that matches business cost, plus residual diagnostics and subgroup performance.
Evaluation · Flagship experience
Regression Metrics & Residuals
What does one error number hide?
Start here
What does one error number hide?
MAE, RMSE and R² summarise different aspects of prediction error. Residual plots reveal structure that a single scalar cannot.
Building interactive view…
Break the assumption deliberately
Add a few large errors and compare how MAE and RMSE respond.
Move the control and explain what you expect before reading the visual.
Formalise what the visual is doing
MAE weights errors linearly; MSE/RMSE square errors; R² compares residual variation with a mean baseline. Heteroscedasticity and systematic residual patterns indicate model mismatch.
Technical questionUse a tiny case to make the mechanism observable. MAE weights errors linearly; MSE/RMSE square errors; R² compares residual variation with a mean baseline. Heteroscedasticity and systematic residual patterns indicate model mismatch. Verify one intermediate quantity, state change or mapping independently; then predict the consequence of this change: Add a few large errors and compare how MAE and RMSE respond.
Use it responsibly
Report a metric that matches business cost, plus residual diagnostics and subgroup performance.
Transfer testUsing accuracy for a rare-event problem without checking class-specific errors.
Worked exploration
Use the visual as an experiment, not decoration
Use residuals -2, 2, -3 and 5. Compute MAE and RMSE. RMSE is larger because squaring gives the residual of 5 disproportionate influence; inspect residual signs/patterns rather than reporting one number alone.
Technical lens
MAE weights errors linearly; MSE/RMSE square errors; R² compares residual variation with a mean baseline. Heteroscedasticity and systematic residual patterns indicate model mismatch.
Practitioner check
Report a metric that matches business cost, plus residual diagnostics and subgroup performance.
Prediction before interaction
Add a few large errors and compare how MAE and RMSE respond.
Exploration walkthrough
Turn the interaction into an evidence trail
Use residuals -2, 2, -3 and 5. Compute MAE and RMSE. RMSE is larger because squaring gives the residual of 5 disproportionate influence; inspect residual signs/patterns rather than reporting one number alone. Before moving the control, state your prediction. After the visual changes, name the specific state, statistic, boundary or mapping that changed and explain why that change is consistent—or inconsistent—with your prediction.
- Record one observable quantity before the interaction and the same quantity afterwards.
- Change one factor at a time so the causal effect of the control is inspectable.
- Use an edge or failure case to discover where the concept stops behaving as the simple story suggests.
Reference depth
Open the complete material
The flagship experience is the map. These pages contain the roads.
Deep Learning Hub lessons
MAE and median absolute errorRegression Metrics & ResidualsMSE and RMSERegression Metrics & ResidualsR² and adjusted R²Regression Metrics & ResidualsResidual analysisRegression Metrics & ResidualsRelated models & simulations
Use the Concept Atlas for related methods and adjacent concepts.
Continue this exact concept
Choose depth, practice or application.
These destinations are explicitly mapped to Regression Metrics & Residuals; they are not generic landing-page fallbacks.