wilson_interval(successes: int, n: int, *, z: float = 1.959963984540054) -> tuple[float, float]
Wilson score interval for a binomial proportion (default 95%).
More reliable than the normal approximation for small n or extreme rates,
and it never leaves [0, 1].
See Also
cohen_kappa : Inter-rater agreement.
flag_rate_ci : Bootstrap interval on audit flag rates.
Examples:
>>> tuple(round(x, 3) for x in wilson_interval(8, 10))
(0.49, 0.943)
Source code in src/gbfs_toolkit/analytics/agreement.py
| def wilson_interval(successes: int, n: int, *, z: float = 1.959963984540054) -> tuple[float, float]:
"""Wilson score interval for a binomial proportion (default 95%).
More reliable than the normal approximation for small ``n`` or extreme rates,
and it never leaves ``[0, 1]``.
See Also
--------
[`cohen_kappa`][gbfs_toolkit.cohen_kappa] : Inter-rater agreement.
[`flag_rate_ci`][gbfs_toolkit.flag_rate_ci] : Bootstrap interval on audit flag rates.
Examples
--------
>>> tuple(round(x, 3) for x in wilson_interval(8, 10))
(0.49, 0.943)
"""
if n == 0:
return (float("nan"), float("nan"))
p = successes / n
z2 = z * z
denom = 1 + z2 / n
center = (p + z2 / (2 * n)) / denom
half = (z * math.sqrt(p * (1 - p) / n + z2 / (4 * n * n))) / denom
return (max(0.0, center - half), min(1.0, center + half))
|