distance_bound: Upper bound for distance between clusters

In aczepielik/CrossTabCluster: Tools For Clsutering Cross Tables

Description

Under the null hypothesis of independence or homogeneity, the maximal squared contrast in rows or columns of contingency table, which can be interpreted as distance between rows is asympotically distributed as the largest eigenvalue of the Wishart matrix W(I_min(I,J), max(I,J)), where I and J are dimensions of the table.

Usage

distance_bound(table, alpha = 0.05, options = list(samples = 1000, seed
  = 42))

Arguments

table	object of class table. The contingency table to calculate the upper bound for.
alpha	double. The returned value is 1-alpha quantile of the underlying distribution. Default is 0.05.
options	list. Two elements taken into accout is "samples", which says how many repetitions to use in simulating distribution (default: 1000) and "seed" used in sampling if not null (default: NULL).

Value

The 1-alpha quantile of the distribution of maximal estimated squared contrast in tables of the size like the given table under the null hypothesis of independence or homogeneity.

Examples

distance_bound(israeli_survey, options = list(samples = 1000, seed = 42))

aczepielik/CrossTabCluster documentation built on June 19, 2020, 7:53 p.m.