plotDominationGraph: Visualize the Pareto fronts of parameter configuration scores

View source: R/visualizations.R

plotDominationGraphR Documentation

Visualize the Pareto fronts of parameter configuration scores

Description

Draws the Pareto fronts and domination relations of tested parameter configurations in a graph. Here, the leftmost column of nodes represents the non-dominated configurations (i.e. the first Pareto front). The second column contains the second Pareto front, i.e. the configurations that are only dominated by the first Pareto front, and so on. An edge between two configurations indicate that the first configuration is dominated by the second.

Usage

plotDominationGraph(tuneParetoResult, 
                    transitiveReduction = TRUE, 
                    drawDominatedObjectives = TRUE, 
                    drawLabels = TRUE, 
                    drawLegend = TRUE,
                    x.legend = "topleft",
                    cex.legend = 0.7,
                    col.indicator, 
                    pch.indicator = 15, 
                    cex.indicator = 0.8,
                    ...)

Arguments

tuneParetoResult

An object of class TuneParetoResult as returned by tunePareto.

transitiveReduction

If this is true, transitive edges in the graph are removed to enhance readability. That is, if configuration c1 dominates configuration c2 and c2 dominates c3, no edge from c3 to c1 is drawn.

drawDominatedObjectives

If set to true, color indicators are drawn next to the nodes. Here, each color corresponds to one objective. The color is drawn next to a node if this node has the best score in this objectives among all solutions of the same Pareto front (i.e., column of the graph).

drawLabels

Specifies whether the parameter configurations should be printed next to the corresponding edges.

drawLegend

If drawDominatedObjectives=TRUE, this specifies whether a legend with the objective colors should be drawn.

x.legend

The position of the legend. For details, refer to the x parameter of legend.

cex.legend

Specifies the size of the text in the legend if drawLegend is true.

col.indicator

Specifies an optional list of colors, one for each objective function. These colors will be used for the indicators if drawDominatedObjectives is true. By default, a predefined set of colors is used.

pch.indicator

Specifies a single plotting character or a list of plotting characters for the objective functions in the indicators which is used for the indicators if drawDominatedObjectives is true.

cex.indicator

Specifies the size of the symbols in the indicators which is be used for the indicators if drawDominatedObjectives is true. This can also be a vector of sizes for the symbols of the objectives.

...

Further graphical parameters for plot.igraph.

Value

Invisibly returns the igraph object representing the graph.

See Also

tunePareto

Examples

# call tunePareto using a k-NN classifier 
# with different 'k' and 'l' on the 'iris' data set
x <- tunePareto(data = iris[, -ncol(iris)], 
                labels = iris[, ncol(iris)],
                classifier = tunePareto.knn(),
                k = c(5,7,9),
                l = c(1,2,3),
                objectiveFunctions=list(cvError(10, 10),
                                        cvSpecificity(10, 10, caseClass="setosa")))

# plot the graph                 
plotDominationGraph(x)

TunePareto documentation built on Oct. 2, 2023, 5:06 p.m.