pod.calculate_reliability_point

calculate_reliability_point(X_eval, ci_lower, target_pod=0.9)

Calculates the defect size (a90/95) where the Lower Confidence Bound crosses the target reliability threshold (usually 0.90).

Parameters

Name	Type	Description	Default
X_eval	np.ndarray	The evaluation grid points.	required
ci_lower	np.ndarray	The lower confidence bound curve (y values).	required
target_pod	float	Target reliability level. Defaults to 0.90.	`0.9`

Returns

Name	Type	Description
float	float	The interpolated x-value where the bound first reaches the target, or
	float	np.nan if it never does. If the bound is already at or above the target at the
	float	first grid point, that point is returned (the true crossing is at or below it)
	float	and a warning is logged.

Examples

import numpy as np
X_eval = np.linspace(0, 10, 101)
lower_ci = 1 / (1 + np.exp(-(X_eval - 5)))
a90_95 = calculate_reliability_point(X_eval, lower_ci, target_pod=0.90)
print(f"a90/95 point: {a90_95:.2f}")