stats

Documentation

Overview

Package stats holds the small deterministic statistics skillet tools need. It is a faithful port of the pure functions in cc-thinking-skills' evals/lib/stats.js. Pure: values in, values out, no I/O.

Index

• func McNemar(improved, regressed int) (stat float64, significant bool)
• func Wilson(k, n int) (lo, hi float64)

Functions

func McNemar

func McNemar(improved, regressed int) (stat float64, significant bool)

McNemar runs McNemar's test over paired before/after binary outcomes: improved is the count that flipped fail→pass, regressed the count that flipped pass→fail. It returns the continuity-corrected χ² statistic and whether it clears the χ²(1, .05) critical value. With no discordant pairs there is no evidence of a difference. Pure.

func Wilson

func Wilson(k, n int) (lo, hi float64)

Wilson returns the Wilson score interval [lo, hi] for k successes out of n Bernoulli trials at 95% confidence. It requires 0 <= k <= n. n == 0 returns the maximally-uncertain [0, 1]; otherwise the bounds lie in [0, 1] and bracket k/n. Pure.