mcframework.stats_engine.kurtosis

mcframework.stats_engine.kurtosis(x: ndarray, ctx: StatsContext) → float

Unbiased sample excess kurtosis (Fisher definition).

\[\text{kurt}(x) = \frac{1}{n_\text{eff}} \sum_i \left(\frac{x_i - \bar X}{s}\right)^4 - 3.\]

Parameters:

xndarray

Input sample.

ctxStatsContext

Uses nan_policy.

Returns:

float

Excess kurtosis (0.0 if \(n_\text{eff} \le 3\)).

Notes

Uses scipy.stats.kurtosis() with fisher=True, bias=False.

Examples

>>> round(kurtosis(np.array([1, 2, 3, 4.0]), {}), 1)
-1.2