icc2_1_estimate

Note

This page is a reference documentation. It only explains the function signature, and not how to use it. Please refer to the Habitat Guide and Python API guide (v2.0) for usage.

icc2_1_estimate(data: ndarray, alpha: float = 0.05) → ICCEstimate[source]

ICC(2,1): two-way RANDOM effects, absolute agreement, single measurement.

Use this for the delineation axis, where the k raters (radiologists or AI segmentations) are a random sample from a rater population and we want the reliability of any single rater, penalising systematic rater bias. The point estimate equals habit.kernels.icc.icc2_1() and pingouin’s ICC(A,1) row; the confidence interval is the Satterthwaite form of McGraw & Wong (1996), not the simpler F interval of the model-3 flavours.

The name carries _estimate to stay distinct from the point-estimate kernel habit.kernels.icc.icc2_1() re-exported alongside it.

Parameters:

  • data – (n, k) matrix, one row per target, one column per rater. NaN handling is the caller’s responsibility.

  • alpha – Two-sided significance level for the confidence limits.

Returns:

The estimate with its (1 - alpha) confidence limits, truncated to [0, 1] per the paper’s convention.

Raises:

ValueError – If data is not a 2-D, NaN-free matrix with at least 2 rows and 2 columns.