emp2pval {bayesclust}R Documentation

Convert Empirical Posterior Probability to P-value

Description

emp2pval converts the Empirical Posterior Probability (EPP) computed by cluster.test into a frequentist p-value, which can then be used to assess the significance of the alternative hypothesis.

Usage

emp2pval(x, y, ignore = FALSE)

Arguments

x x is an object of class ``cluster.test'', which is returned when cluster.test is run on a dataset.
y y is an object of class ``nulldensity'', which is returned when nulldensity is run with the necessary parameters provided.
ignore ignore is a logical variable which specifies whether the parameters in x and y should be matched for consistency.

Details

If ignore is set to FALSE, then the routine will first check to see if the parameters under which the test was run match exactly with the parameters under which the null distribution was generated. If they were, then the EPP's in the ``cluster.test'' object will be converted to a frequentist p-value by checking the ``nulldensity'' object to see which empirical quantile they fall in. If ignore is set to TRUE, the same EPP to p-value conversion is carried out, but this time without the preliminary check on the parameters.

When several ``emp2pval'' objects are created, i.e., when several datasets are to be tested or when multiple tests are carried out on a single dataset, they can all be fed into fdr.test to assess which tests are significant, while controlling the False Discovery Rate (FDR).

Value

emp2pval returns a list comprising 2 components.

param This component is a copy of the parameters used when running cluster.test on the dataset. In the case that emp2pval was run with ignore set to TRUE, then this could potentially be different than the parameters under which the nulldensity object was generated.
pvals A dataframe with the EPP of the dataset and the corresponding frequentist p-values.

Author(s)

Gopal, V.

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

See Also

cluster.test for information on objects of class ``cluster.test''.

nulldensity for information on objects of class ``nulldensity''.

Examples

# Generate random 2-variate data
Y <- matrix(rnorm(24), nrow=12)

# Search for optimal partitioning of data into 2 clusters
test1 <- cluster.test(Y, p=2)

# Generate corresponding null density object.
null1 <- nulldensity(nsim=100, n=12, p=2, k=2)

# Convert EPP to p-value
emp2pval(test1, null1)


[Package bayesclust version 2.1 Index]