SuperGLM

SuperGLM is the estimator for penalised GLMs and GAM-style pricing models. Construct it with a family and a feature specification, fit it with fit_reml() for REML smoothness selection or fit() for fixed penalties, then read the fit through summary(), term_inference() and the plotting methods. The strip below follows a model through its life. Each page under it opens with the members you reach for first, and its tables list every member in the group, each with its own page.

superglm.SuperGLM

Penalised generalised linear model with splines, group penalties, and REML.

1 · Build

A family, a feature spec, a penalty policy.

Build

2 · Fit

fit_reml for REML; fit for a fixed penalty.

Fit

3 · Inference

Summary, per-term curves, diagnostics.

Inference

4 · Predict

Means, relativities, reconstructed effects.

Predict

5 · Deploy

Rating tables and the payload behind them.

Deploy

All the groups, in the order you meet them:

• Build
• Fit
• Inference
• Predict
• Plot
• Diagnose
• Constrain shapes
• Screen interactions
• Deploy
• Results and records