dkwSampleSize

fun dkwSampleSize(eta: Double, alpha: Double = 0.05): Double(source)

The Dvoretzky–Kiefer–Wolfowitz sample size: how many observations are needed for the empirical distribution function to lie within a given distance of the truth, uniformly, with the requested confidence.

Present for comparison rather than as part of the bound. It answers a different and easier question — how much data represents the distribution — and the contrast between the two is the point being made.

Parameters

eta

the uniform accuracy required of the empirical distribution function

alpha

one minus the confidence