deltaplot {TopKLists}

An exploratory plot of discordance for delta selection.

Description

Returns a graph of non-overlap (discordance) of rankings represented by the sum of zeros across all objects in the \delta-dependent Idata vector (see compute.stream) for a suitable range of \delta values starting at \delta=0. Graphs are plotted for all pairwise list combinations.

Usage

deltaplot(lists, deltas=NULL, subset.lists=NULL, subplot = FALSE,
perc.subplot=50, directory=NULL) 

Arguments

lists

A data frame cotaining two or more columns that represent lists of ordered objects to be compared

deltas

The range of \delta values to be examined, defaults to NULL. If not specified then delta=c(1:nrow(lists)*0.25). If max(deltas) is larger than nrow(lists), then deltas=deltas[which(deltas<nrow(lists)*0.25)],

subset.lists

Specifies the subset of the input lists, which is used for calculating zero counts for the deltaplot. The value contained in subset.lists specifies which objects are taken for the calculation from each input list, e.g. a value of 100 would use the first 100 objects of each input list. Default is NULL, in which case all objects of each list are used. If specified and max(deltas) is larger than subset.lists, then deltas=deltas[which(deltas<subset.lists*0.25)],

subplot

Logical: if TRUE an additional deltaplot is generated containing a detailed subplot positioned in the top right corner of the plot. This subplot encloses a configurable subset of the values of the original deltaplot. This subset can be specified via a percentage value using the perc.subplot parameter (default is FALSE for subplot).

perc.subplot

Percentage of the range of the main plot used for creating a subplot in the top right corner, default is 50(%). Subplot provides a detailed view of the main plot.

directory

Specifies the directory for saving the generated deltaplots in PDF format. In case directory is NULL (default), no pdf is created, but a new device will be opened.

Value

Mdelta

A list of \delta-matrices for each comparison of ordered lists

Author(s)

Eva Budinska <[e-mail]>, Vendula Svendova <[e-mail]>, Michael G. Schimek <[e-mail]>

References

Schimek, M. G. and Budinska, E. (2010). Visualization techniques for the integration of rank data. In Lechevallier, Y. and Saporta, G. (eds). COMPSTAT 2010. Proceedings in Computational Statistics. Heidelberg: Physica (e-book ISBN 978-3-7908-2603-6), 1637-1644.

Examples

set.seed(1234)
data(breast)
##plot subplot
a = deltaplot(breast, deltas = 1:50, subplot=TRUE)

##don't plot subplot (default)
a = deltaplot(breast, deltas=1:50, subplot = FALSE)