plotimp: Replot attribute relevances with different plot parameters...

In mkampert/rCOSA: Clustering Objects on Subsets of Attributes (COSA)

Description

This function allows replotting of the attribute revelances (importances) with different display parameters after calling "attimp", without having to recompute the relevance values. It also allows flexibility in how the results are displayed.

Usage

plotimp(
  attimp,
  atts = attimp$att[1:min(40, length(attimp$att))],
  imporder = F,
  donames = T,
  trans = 1,
  horiz = F,
  main = "Cluster",
  att.names = 1:length(attimp$att),
  names.size = 1,
  plot = T,
  ylim = c(0, max(imps)),
  ntick = 20
)

Arguments

attimp	output from "attimp"
atts	vector of attribute identities. This vector can be numeric, referencing the respective attribute numbers, or character, referencing the respective attribute names. If the latter, then the parameter "att.names" (see below) must be supplied.
imporder	TRUE / FALSE => do/don't plot relevance values in order of descending value.
donames	TRUE / FALSE => do/don't show names of attributes on plot.
trans	power used to transform importance scale.
horiz	TRUE / FALSE => do/don't plot importances horizontally
main	plot title
att.names	vector of names for each attribute (ignored if donames= FALSE).
names.size	expansion factor for names.
plot	TRUE / FALSE => do/don't display barplot.
ylim	vertical range of plot.
ntick	number of tick marks on importance axis.

Value

values of attribute importances in requested order.

See Also

attimp

Examples

# make sure an at object is available...
# plotimp(at)
# plotimp(at,1:100,trans=0.5,horiz=TRUE)
# plotimp(at,c(27,43,10,16))
# plotimp(at,c('age','income','occupation'),att.names=demographics)

mkampert/rCOSA documentation built on Dec. 23, 2019, 8:21 p.m.