cutoffs {bayesclust}R Documentation

Table of Cut-Off Points

Description

This is a table of pre-computed cut-off points for testing significance of clusters at the alpha=0.05 and 0.01 level.

Usage

data(cutoffs)

Format

A table of cut-off points obtained by generating the null distribution of the posterior probability using the following parameters:

n
The number of observations in the dataset.
mcs
mcs stands for Minimum Cluster Size.
p
The length of the vector of each observation. For example, p=2 corresponds to bivariate data.
k
The precise (simple) alternative hypothesis being tested.
cutoff1pct
A numeric vector consisting of the cut-off points for the alpha=0.01 level.
cutoff5pct
A numeric vector consisting of the cut-off points for the alpha=0.05 level.

Details

In order to test the significance of the Empirical Posterior Probability (EPP), it is necessary to generate its distribution under H_0, and compare the EPP to the sample quantile for the desired level of significance. However this simulation could take a considerably long time, and hence this table is provided to enable the experimenter to get a crude estimate of critical values at the alpha=0.05 and alpha=0.01 levels. The parameters under which the table was generated are part of the dataframe, allowing the experimenter to choose the closest set of conditions to his/her particular set-up.

References

Fuentes, C. and Casella, G. (2008) "Testing for the Existence of Clusters" http://www.stat.ufl.edu/~casella/Papers/paper-v3.pdf

Gopal, V. "BayesClust User Manual" http://www.stat.ufl.edu/~viknesh/bayesclust/clust.html

Examples

data(cutoffs)
## maybe str(cutoffs) ; plot(cutoffs) ...

[Package bayesclust version 2.1 Index]