Bound

data class Bound(val label: String, val hellinger: Double, val kolmogorov: Double, val mergedPair: Int, val mergedFamily: String, val mergedComponent: ContinuousDistributionIfc)(source)

What the search found for one case.

Parameters

label

the case label

hellinger

the smallest Hellinger distance to a merged (k-1)-component mixture

kolmogorov

the largest gap between the two distribution functions at that merge

mergedPair

the index of the lower component of the pair whose merge was closest

mergedFamily

the family the merged component took

mergedComponent

the single distribution that replaced the pair. Carried so that the approximating mixture can be drawn, not merely scored: a figure of the gap has to plot the density that produced the number

Constructors

Link copied to clipboard
constructor(label: String, hellinger: Double, kolmogorov: Double, mergedPair: Int, mergedFamily: String, mergedComponent: ContinuousDistributionIfc)

Properties

Link copied to clipboard
Link copied to clipboard

The sample size below which no procedure can separate the two models with bounded error, of the order of one over the squared Hellinger distance.

Link copied to clipboard
Link copied to clipboard
Link copied to clipboard
Link copied to clipboard