map.convergence: SOM Quality Assessment

In lutzhamel/popsom2: Functions for Constructing and Evaluating Self-Organizing Maps

Description

Evaluate the quality of a SOM using embedding accuracy and estimated topographical accuracy.

Usage

map.convergence(map,conf.int=.95,k=50,verb=FALSE,ks=FALSE)

Arguments

map	an object of type 'map'.
conf.int	is the confidence interval of the quality assessment (default 95%)
k	number of samples to use in the computation of the estimated topographical accuracy (default=50)
verb	if true reports the two convergence components separately, otherwise it will report a linear combination of the two indices (default=FALSE)
ks	if true uses the Kolmogorov-Smirnov convergence test otherwise a convergence test based on variance and means is performed (default=FALSE)

Value

A single value or a pair of values: 1) embedding accuracy 2) estimated topographic accuracy.

Author(s)

Lutz Hamel

References

"SOM Quality Measures: A Statistical Approach," Lutz Hamel, WSOM16, 2016.

Examples

data(iris)

## set data frame and labels
df <- subset(iris,select=-Species)
labels <- subset(iris,select=Species)

## build a map
m <- map.build(df, labels, xdim=15, ydim=10, train=1000)

## map quality
map.convergence(m)

lutzhamel/popsom2 documentation built on May 25, 2020, 12:46 a.m.