pymargins.DiagnosticResult

class pymargins.DiagnosticResult(kappa_min: float, kappa_median: float, kappa_max: float, kappa_distribution: ndarray, verdict: str, n_samples: int, recommendation: str, session_summary: str = '')

Output of session-level kappa diagnostic.

Returned by Margins.diagnose() to summarize delta-method validity across the design space before any specific estimand is computed.

kappa_min, kappa_median, kappa_max

Summary stats of κ across sampled covariate vectors.

Type:

float

kappa_distribution

All sampled κ values, for inspection.

Type:

array

verdict

Classification: “delta_reliable”, “delta_borderline”, or “delta_unreliable”, driven by max κ vs configured thresholds.

Type:

str

n_samples

How many design points were sampled.

Type:

int

recommendation

Human-readable advice based on verdict.

Type:

str

session_summary

One-line summary of the session’s analytical posture (scale, vcov, method) for context in audit logs.

Type:

str

__init__(kappa_min: float, kappa_median: float, kappa_max: float, kappa_distribution: ndarray, verdict: str, n_samples: int, recommendation: str, session_summary: str = '') → None

Methods

__init__(kappa_min, kappa_median, kappa_max, ...)
summary()

Attributes

session_summary
kappa_min
kappa_median
kappa_max
kappa_distribution
verdict
n_samples
recommendation