randIndexDetailed: Compute Rand Contingency Index from a contingency table.

In jvanheld/stats4bioinfo: Utilities for the book "Statistics for bioinformatics"

Description

Compute Rand Contingency Index from a contingency table..

Usage

randIndexDetailed(cont.table, permute.cols = FALSE, permute.rows = FALSE)

Arguments

cont.table	a contingency table (supposed to contain counts). Can be produced by table() or other means. Can belong to any of the following classes: table, matrix or data.frame.
permute.cols=FALSE	permute columns of the contingency table by decreasing sum.
permute.rows=FALSE	permute rows of the contingency table by decreasing sum.
...	Additional parameters are passed to the function image()

Details

First version: 2016-12-26 Last modification: 2016-12-26

The Rand Consistency Index is a classical measure of the similarity between two classifications (e.g. clustering results). It relies on the number of consistent pairs, where "consistent" means either co-clusttered in both classifications or separated in both.

Value

a list with the values of the computed parameters.

Author(s)

Jacques van Helden ([e-mail])

Examples

clustering1 <- c(rep(c("a", "b","c"), times=c(4,2,1)))
clustering2 <- c(rep(c("X", "Y"), times=c(3,4)))
cont.table <- table(clustering1, clustering2)

## Example from Yeung and Ruzzo (2001) supplementary file
cont.table <- data.frame(v1 = c(1, 1, 0), v2 = c(1, 2, 0), v3=c(0, 1, 4), row.names=c("u1", "u2", "u3"))
result <- randIndexDetailed(cont.table)
print(result$consistency.table)
print(signif(unlist(result[c("n", "Npairs", "RI", "ARI")]), digits=3))

## Check the result with another implementation
flexclust::randIndex(as.table(as.matrix(cont.table)), correct=TRUE)

## Draw heatmaps of the contingency tables
heatmap.simple(cont.table)
heatmap.simple(result$cont.table)

jvanheld/stats4bioinfo documentation built on May 20, 2019, 5:16 a.m.