dispersion_test

superglm.dispersion_test(

model: SuperGLM,

X,

y,

sample_weight=None,

offset=None,

*,

alternative: str = 'greater',

) → DispersionTestResult

Cameron & Trivedi (1990) regression-based overdispersion test.

Tests H0: Var(Y) = mu (equidispersion) vs H1: Var(Y) = mu + alpha*mu^2.

Parameters:

modelSuperGLM

A fitted Poisson model.

Xpandas or eager Polars DataFrame

Design matrix.

yarray-like

Observed response values.

sample_weightarray-like or None

Observation weights. Currently unsupported.

offsetarray-like or None

Optional offset.

alternativestr

One of "greater" (overdispersion), "less" (underdispersion), or "two-sided".

Returns:

DispersionTestResult