plot.somtune: Plot SOM parameter tuning results

In yasomi: Yet Another Self Organising Map Implementation

Description

Visualise the results of parameter tuning for the SOM

Usage

## S3 method for class 'somtune'
plot(x, relative = TRUE,
sqrt.quant=!x$isquant, best.color="red", xlab, ylab, yaxs, legend.text, ...)

Arguments

x	an object of class somtune
relative	specifies whether the best (i.e., lowest) value of the criterion should be used as a baseline in the display (the default behaviour) or not
sqrt.quant	specifies whether the plot should display the quantisation error or its square root. When som.tune has been used with the quantisation error as the quality measure, the default is to show it directly, while is square root is used in other cases
best.color	gives the colour of the border of bar associated to the best configuration. Use NA to suppress this effect
xlab	a label for the x axis (see barplot)
ylab	a label for the y axis (see barplot)
yaxs	the style of axis interval calculation to be used for the y-axis (see par)
legend.text	a vector of text used to construct a legend for the plot (see barplot)
...	not used

Details

This function produces a barplot with the quantisation error of all the tested SOMs as well as an additional bar for an another error criterion if the quantisation error was not used to choose the best SOM. Most of the parameters have reasonable default values computed from the somtune object. The function tries to avoid superposition of the legend with the bars.

Warning

This function works only when one parameter is tuned by som.tune. It will not work in particular if random initialisation is used several times together with a parameter tuning.

Author(s)

Fabrice Rossi

See Also

som.tune and som.tunecontrol

Examples

data(iris)
# scaling
data <- scale(iris[1:4])

# a medium hexagonal grid
sg <- somgrid(xdim=15,ydim=15,topo="hex")

# choose a good SOM via Kaski and Lagus' error measure
tune.results <-
  som.tune(data,sg,som.tunecontrol(sg,criterion=error.kaskilagus))

# plot errors
plot(tune.results)

yasomi documentation built on May 2, 2019, 5:59 p.m.