wilson-interval

A comprehensive module used to calculate the high bound, low bound, and center of a Wilson score interval.

various confidence interval implemented in javascript

能力衰变哨兵 DecayWatch — 长 agent 运行中按能力类目基线成功率并预警中途衰变

Wilson confidence intervals, difference of proportions, and McNemar's test for LLM evaluation results. Client-side, single file.

Domain-pluggable engine for competitive-play analysis — games plug in via a six-point contract: percentiles, Wilson intervals, meta drift. Adapters: Dota 2, chess.