Skip to content

huge_inference

Usage

huge_inference(
    data,
    t,
    adj,
    alpha=0.05,
    type_="Gaussian",
    method="score",
) -> HugeInferenceResult

Description

Native edge-wise inference helper using a partial-correlation z-test approximation.

Key arguments

  • data: sample matrix (n x d) with n >= 2 and no constant columns; Nonparanormal inference also requires d >= 2
  • t: finite precision-like matrix (d x d) with a positive diagonal
  • adj: reference adjacency (d x d)
  • type_: "Gaussian" or "Nonparanormal"
  • method: "score" or "wald" for Nonparanormal inference; ignored for Gaussian inference

Returns

HugeInferenceResult with transformed data, p-value matrix p, and error. Numerically degenerate inputs that produce non-finite edge p-values raise PyHugeError instead of returning an apparently successful result.