Plot

plot() is the single entry point for drawing terms: all main effects, one, a subset, or an interaction, with pointwise or simultaneous bands on the main effects. plot_data() returns the plain DataFrames, arrays and metadata behind those figures so you can rebuild them in matplotlib, plotly, Excel or a reporting system. plot_diagnostics() is the four-panel residual figure on quantile residuals, with a simulation-based Q-Q envelope.

plot	Plot model terms.
plot_data	Return plain data needed to recreate SuperGLM plots.
plot_diagnostics	GLM/GAM diagnostic figure with simulation-based Q-Q envelope.

Example

The same simulated motor book as the rest of this section: 4,000 policies, a non-linear age effect, four regions and a mild vehicle-power slope, with exposure carried as a log offset.

import numpy as np
import pandas as pd

from superglm import Categorical, Numeric, Spline, SuperGLM

rng = np.random.default_rng(0)
n = 4000
region = rng.choice(["North", "South", "East", "West"], n, p=[0.35, 0.3, 0.2, 0.15])
age = rng.uniform(18, 80, n)
veh_power = rng.uniform(4, 12, n)
exposure = rng.uniform(0.1, 1.0, n)
region_effect = pd.Series(region).map(
    {"North": 0.0, "South": 0.25, "East": -0.2, "West": 0.4}
).to_numpy()
log_rate = (
    -1.2
    + 1.1 * np.exp(-((age - 24) ** 2) / 90.0)
    + 0.004 * (age - 50) ** 2 / 10.0
    + region_effect
    + 0.09 * (veh_power - 8)
)
claims = rng.poisson(np.exp(log_rate) * exposure)
X = pd.DataFrame({"age": age, "region": region, "veh_power": veh_power})
offset = np.log(exposure)

model = SuperGLM(
    family="poisson",
    features={
        "age": Spline(kind="cr", k=10),
        "region": Categorical(),
        "veh_power": Numeric(),
    },
).fit_reml(X, claims, offset=offset)
model.reml_diagnostics()["converged"]

True

One named term draws one figure: the fitted age curve with its pointwise band and, because X was passed, the density of the fitting rows underneath, so a thin stretch of data cannot masquerade as a confident part of the curve. Drop the term name to draw every main effect, pass a list for a subset, and pass ci="simultaneous" for bands that hold jointly across the curve.

fig = model.plot("age", X=X, engine="matplotlib")

plot_data returns the numbers behind that figure instead of drawing it: one entry per term, each with the effect frame, the density, and the metadata describing how the curve was centred and how much of the basis the fit used. Nothing in it needs SuperGLM to render — this is the handover to plotly, Excel or a reporting system.

payload = model.plot_data("age", X=X)
age_term = payload["terms"][0]
print(
    f"kind={payload['kind']}, centering={age_term['metadata']['centering_mode']}, "
    f"edf={age_term['metadata']['edf']:.2f}"
)
age_term["effect"].head().round(3)

kind=main_effects, centering=training_mean_zero_unweighted, edf=4.58

	x	log_relativity	relativity	se_log_relativity	ci_lower	ci_upper
0	18.007	0.988	2.687	0.093	2.240	3.223
1	18.318	0.981	2.668	0.088	2.244	3.171
2	18.630	0.974	2.649	0.084	2.248	3.121
3	18.941	0.967	2.630	0.079	2.251	3.073
4	19.253	0.960	2.611	0.075	2.252	3.026

plot_diagnostics is the residual view rather than the effect view: four panels on quantile residuals, with the Q-Q panel carrying an envelope simulated from the fitted model, so the question “is this tail worse than this model would produce anyway” has an answer on the page. It needs the frame and the response back, because residuals are not stored on the model.

diagnostics = model.plot_diagnostics(X, claims, offset=offset)