Data Visualization
Predictor vs Response
Observed
Fitted
Mean/Probability
Model Fit
Predicted vs Actual
Fit the model to see calibration.
Diagnostics
Residual Plot
Choose raw or deviance residuals.
Link Transformation
η and μ Side by Side
Inverse link maps the linear predictor back to the mean.
Model Comparison
Compare Current Link with an Alternate Fit
This is especially useful for seeing why the canonical link often behaves better.
Generate data and fit the main model first, then compare with a second link on the same dataset.
Deviance
--
AIC
--
Mean Response
--
Residual Spread
--
Diagnostics Panel
Coefficient Table
| Term | Estimate | Std. Error | Z / t | Interpretation |
|---|---|---|---|---|
| Fit a model to populate estimates. | ||||
Learning Mode
What Is Happening?
Teaching mode is on.
Why a link function?
The link lets a linear predictor control a mean that must obey the distribution's constraints.
Variance pattern
Different GLM families change how variance grows with the mean response.
Coefficient meaning
Interpret slope coefficients on the link scale first, then map back to the response scale.
IRLS update
IRLS repeatedly solves a weighted least squares problem until the estimates stabilize.