scatterbox: Plot scatterplots of data with boxes shown

In sdtoolkit: Scenario Discovery Tools to Support Robust Decision Making

Description

Function for plotting scatterplots of cases, with points color-coded by relevance and with two-dimensional projections of scenarios-as-boxes.

Usage

scatterbox(boxseq, boxnum = 1, xdim, ydim, filterdims = FALSE, 
  filterothers = FALSE, nobox = FALSE, addbox = FALSE, 
  data = attr(boxseq, "data"), y = NULL, manualbox = FALSE, ...)

Arguments

boxseq	A box sequence object as ouput by sdprim. Alternatively, a manually created list of box definitions (see details.
boxnum	Integer - Which box in the box sequence should be plotted?
xdim	Integer or Character - Which dimension should be plotted as the x-axis? The safest way is to specify the dimension name. See details.
ydim	Integer or Character - Which dimension should be plotted as the y-axis? The safest way is to specify the dimension name. See details.
filterdims	Logical - If a box is defined by more than just the two dimensions that are being plotted, should points filtered out by the other dimensions be removed from the scatter plot? See details.
filterothers	Specifies whether and how points captured by other boxes should be filtered out. Either TRUE, FALSE, “PREV” or a vector of numerics. TRUE removes points captured by all other boxes. FALSE removes none. “PREV” removes any boxes that came earlier in the covering sequence. Alternatively, a vector of integers may be specified indicating which boxes should have their points removed. Note that in all cases, boxes may overlap and specifying this option may result in not all points in the current box being shown.
nobox	Logical - should the scatterplot be generated, but with no box plotted?
addbox	Logical - should a different box be plotted on the same scatterplot? Note that depending on how points were filtered and which dimensions are involved, this may or may not be misleading.
data	Matrix - the data to use in generating the scatterplots. The default is to use the data attribute returned by sdprim. If this is not used for some reason (such as training versus test data), the dataset should have the same number of dimensions, same dimension names, and same dimension ordering as that used to generate the boxes.
y	Vector of zeros and ones indicating interestingness or uninterestingness of points. If left as NULL, it will assume the item data above is of the form returned by sdprim.
manualbox	Logical - Is the boxseq argument given a manually generated box list not conforming to the output of sdprim? This may be handy if you want to plot boxes or box sequences that you specified yourself independent of sdprim. See details for how to construct such a box.
...	Arguments that may be passed on to pbox and colptplot.

Value

Nothing returned, used for its plotting capabilities.

Author(s)

Benjamin P. Bryant, [e-mail]

References

RAND Pardee Center: http://www.rand.org/international\_programs/pardee/

Examples

data(exboxes)
scatterbox(exboxes,xdim="stations",ydim="depth") 

Example output

sdtoolkit documentation built on May 2, 2019, 6:04 a.m.