Plot absolute and relative information of influence curves.
infoPlot(object, ...)
# S4 method for IC
infoPlot(object, data = NULL,
..., withSweave = getdistrOption("withSweave"),
col = par("col"), lwd = par("lwd"), lty,
colI = grey(0.5), lwdI = 0.7*par("lwd"), ltyI = "dotted",
main = FALSE, inner = TRUE, sub = FALSE,
col.inner = par("col.main"), cex.inner = 0.8,
bmar = par("mar")[1], tmar = par("mar")[3],
with.automatic.grid = TRUE,
with.legend = TRUE, legend = NULL, legend.bg = "white",
legend.location = "bottomright", legend.cex = 0.8,
x.vec = NULL, scaleX = FALSE, scaleX.fct, scaleX.inv,
scaleY = FALSE, scaleY.fct = pnorm, scaleY.inv=qnorm,
scaleN = 9, x.ticks = NULL, y.ticks = NULL,
mfColRow = TRUE, to.draw.arg = NULL,
cex.pts = 1, cex.pts.fun = NULL, col.pts = par("col"),
pch.pts = 19,
cex.npts = 1, cex.npts.fun = NULL, col.npts = grey(.5),
pch.npts = 20,
jitter.fac = 1, with.lab = FALSE, cex.lbs = 1, adj.lbs = c(0, 0),
col.lbs = col.pts, lab.pts = NULL, lab.font = NULL, alpha.trsp = NA,
which.lbs = NULL, which.Order = NULL, which.nonlbs = NULL,
attr.pre = FALSE, return.Order = FALSE,
ylab.abs = "absolute information",
ylab.rel= "relative information",
withSubst = TRUE)
An S3 object of class c("plotInfo","DiagnInfo")
, i.e., a list
containing the information needed to produce the
respective plot, which at a later stage could be used by different
graphic engines (like, e.g. ggplot
) to produce the plot
in a different framework. A more detailed description will follow in
a subsequent version.
object of class "InfluenceCurve"
optional data argument --- for plotting observations into the plot;
logical: if TRUE
(for working with Sweave
)
no extra device is opened
logical: is a main title to be used? or
just as argument main
in plot.default
.
logical: do panels have their own titles? or
character vector of / cast to length 'number of compared
dimensions';
if argument to.draw.arg
is used, this refers to
a vector of length 1
(absolute information)
+ length(to.draw.arg)
, the actually plotted
relative informations. For further information, see also
main
in plot.default
.
logical: is a sub-title to be used? or
just as argument sub
in plot.default
.
top margin -- useful for non-standard main title sizes; may be a vector with individual values for each of the panels to be plotted.
bottom margin -- useful for non-standard sub title sizes; may be a vector with individual values for each of the panels to be plotted.
color of IC in argument object
.
linewidth of IC in argument object
.
line-type of IC in argument object
.
color of the classically optimal IC.
linewidth of the classically optimal IC.
line-type of the classically optimal IC.
magnification to be used for inner titles relative
to the current setting of cex
; as in
par
.
character or integer code; color for the inner title
logical; should a grid be plotted alongside
with the ticks of the axes, automatically? If TRUE
a respective
call to grid
in argument panel.first
is ignored.
logical; shall a legend be plotted?
either NULL
or a list of length (number of plotted panels)
of items which can be used as argument legend
in
command legend
.
a valid argument x
for legend
---
the place where to put the legend on the last issued
plot --- or a list of length (number of plotted panels)
of such arguments, one for each plotted panel.
background color for the legend
magnification factor for the legend
a numeric vector of grid points to evaluate the influence curve;
by default, x.vec
is NULL
; then the grid is
produced automatically according to the distribution of the IC.
x.vec
can be useful for usage with a rescaling of the
x-axis to avoid that the evaluation points be selected too
unevenly (i.e. on an equally spaced grid in the original scale,
but then, after rescaling non-equally).
The grid has to be specified in original scale; i.e.; when used
with rescaling, it should be chosen non-equally spaced.
logical; shall X-axis be rescaled (by default according to the cdf of the underlying distribution)?
logical; shall Y-axis be rescaled for abs.info-plot (by default according to a probit scale)?
an isotone, vectorized function mapping the domain of the IC
to [0,1]; if scaleX
is TRUE
and scaleX.fct
is
missing, the cdf of the underlying observation distribution.
the inverse function to scale.fct
, i.e., an isotone,
vectorized function mapping [0,1] to the domain of the IC
such that for any x
in the domain,
scaleX.inv(scaleX.fct(x))==x
; if scaleX
is TRUE
and scaleX.inv
is
missing, the quantile function of the underlying observation distribution.
an isotone, vectorized function mapping the range of the norm of the IC to [0,1]; defaulting to the cdf of \({\cal N}(0,1)\); can also be a list of functions with one list element for each of the panels to be plot.
an isotone, vectorized function mapping [0,1] into the range of the norm of the IC; defaulting to the quantile function of \({\cal N}(0,1)\); can also be a list of functions with one list element for each of the panels to be plot.
integer; defaults to 9; on rescaled axes, number of x and y ticks if drawn automatically;
numeric; defaults to NULL; (then ticks are chosen automatically); if non-NULL, user-given x-ticks (on original scale);
numeric; defaults to NULL; (then ticks are chosen automatically); if non-NULL, user-given y-ticks (on original scale); can be a list with one (numeric or NULL) item per panel
shall default partition in panels be used --- defaults to TRUE
Either NULL
(default;
everything is plotted) or a vector making a selection
among the relative information plots; the absolute
information being plotted in any case. This
vector is either a vector of integers
(the indices of the subplots to be drawn) or characters
--- the names of the subplots to be drawn: these
names are to be chosen either among the row names of
the trafo matrix
rownames(trafo(eval(object@CallL2Fam)@param))
or if the last expression is NULL
a
vector "dim<dimnr>"
, dimnr
running through
the number of rows of the trafo matrix.
logical; if TRUE
(default) pattern substitution for
titles and lables is used; otherwise no substitution is used.
color of the points of the data
argument plotted;
can be a vector or a matrix. More specifically, if argument attr.pre
is TRUE
, it is recycled to fill a matrix of dimension n
by 2
(n
the number of observations prior to any selection) where filling
is done in order column first. The two columns are used for possibly
different colors for the actual IC from the argument and the classical
IC which is also shown. The selection done via which.lbs
and
which.Order
is then done afterwards and on this matrix;
argument col.npts
is ignored in this case. If attr.pre
is FALSE
,
col.pts
is recycled to fill a matrix of dimension n.s
by 2
where n.s
is the number of observations selected for labelling
and refers to the index ordering after the selection. Then argument
col.npts
deteremines the colors of the shown but non-labelled
observations as given in argument which.nonlbs
.
symbol of the points of the data
argument plotted
(may be a vector of length 2 or a matrix, see col.pts
,
with argument pch.npts
as counterpart).
size of the points of the data
argument plotted
(may be a vector of length 2 or a matrix, see col.pts
,
with argument cex.npts
as counterpart).
rescaling function for the size of the points to be plotted;
either NULL
(default), then log(1+abs(x))
is used for each of
the rescalings, or a function which is then used for each of the
rescalings, or a list of functions; if it is a function or a list of
functions, if necessary it is recylced to length 2 * dim
where 2
is for the classical IC and the IC in argument object
and dim
is the number of dimensions of the pICs to be plotted;
in the index of this list, 2
is incremented first;
then dim
.
color of the non-labelled points of the data
argument
plotted; (may be a vector of length 2, or it can be a matrix
nnlb <- sum(which.nonlbs)
by 2
,
nnlb
the number of non-labelled shown observations.
symbol of the non-labelled points of the data
argument
plotted (may be a vector of length 2
or a matrix,
see col.npts
).
size of the non-labelled points of the data
argument
plotted (may be a vector of length 2
or a matrix, see
col.npts
).
rescaling function for the size of the non-labelled points
to be plotted; either NULL
(default), then log(1+abs(x))
is used for each of the rescalings, or a function which is then used
for each of the rescalings, or a list of functions; if it is a
function or a list of functions, if necessary it is recylced
to length 2 * dim
where dim
is the number of dimensions of
the pICs to be plotted; in the index of this list,
2
is incremented first; then dim
.
logical; do graphical attributes for plotted data refer
to indices prior (TRUE
) or posterior to selection
via arguments which.lbs
, which.Order
, which.nonlbs
(FALSE
)?
logical; shall labels be plotted to the observations?
(may be a vector of length 2, see col.pts
--
but not a matrix)
size of the labels; can be vectorized to an array
of dim nlbs x 2 x npnl where npnl is the number of plotted
panels and nlbs the number of plotted labels; if it is
a vector, it is recylced in order labels then ICs
[arg IC
/classic] then panels.
color of the labels; can be vectorized to a matrix
of dim nlbs x 2 as col.pts
.
adjustment of the labels; can be vectorized to an array
of dim 2 x 2 x npnl matrix, npnl the number of plotted panels;
if it is a vector, it is recycled in order (x,y)-coords then ICs
[arg IC
/classic] then panels.
character or NULL; labels to be plotted to the observations;
can be a vector of length n
, n
the number of
all observations prior to any selection with which.lbs
,
which.Order
; if lab.pts
is NULL
,
observation indices are used.
font to be used for labels; (may be a vector of length 2,
see with.lab
).
alpha transparency to be added ex post to colors
col.pch
and col.nonlbl
; if one-dim and NA all colors are
left unchanged. Otherwise, with usual recycling rules alpha.trsp
gets shorted/prolongated to length the number of panel data-symbols to
be plotted. Coordinates of this vector alpha.trsp
with NA are left unchanged,
while for the remaining ones, the alpha channel in rgb space is set
to the respective coordinate value of alpha.trsp
. The non-NA
entries must be integers in [0,255] (0 invisible, 255 opaque).
jittering factor used in case of a DiscreteDistribution
for plotting points of the data
argument in a jittered fashion
(may be a vector of length 2, see with.lab
).
either an integer vector with the indices of the observations
to be plotted into graph or NULL
--- then no observation is excluded
we order the observations (descending) according to the norm given by
normtype(object)
; then which.Order
either is an integer vector with the indices of the ordered
observations (remaining after a possible reduction by argument which.lbs
)
to be plotted into graph or NULL
--- then no (further) observation
is excluded.
indices of the observations which should be plotted but
not labelled; either an integer vector with the indices of the observations
to be plotted into graph or NULL
--- then all non-labelled
observations are plotted.
logical; if TRUE
, a list of length two with order vectors
is returned --- one for ordering w.r.t. the given IC, one for ordering w.r.t.
the classically optimal IC; more specifically, the order of the (remaining) observations
given by their original index is returned (remaining means: after a possible
reduction by argument which.lbs
, and ordering is according to the norm given by
normtype(object)
);
otherwise we return invisible()
as usual.
character; label to be used for y-axis in absolute information panel
character; label to be used for y-axis in relative information panel
further parameters for plot
Matthias Kohl Matthias.Kohl@stamats.de
Absolute information is defined as the square of the length
of an IC. The relative information is defined as the
absolute information of one component with respect to the
absolute information of the whole IC; confer Section 8.1
of Kohl (2005).
Any parameters of plot.default
may be passed on to this particular
plot
method.
For main-, inner, and subtitles given as arguments main
,
inner
, and sub
, top and bottom margins are enlarged to 5 resp.
6 by default but may also be specified by tmar
/ bmar
arguments.
If main
/ inner
/ sub
are
logical then if the respective argument is FALSE
nothing is done/plotted,
but if it is TRUE
, we use a default main title taking up the calling
arguments in case of main
, default inner titles taking up the
class and (named) parameter slots of arguments in case of inner
,
and a "generated on <data>"-tag in case of sub
.
Of course, if main
/ inner
/ sub
are character
, this
is used for the title; in case of inner
it is then checked whether it
has correct length. If argument withSubst
is TRUE
, in all title
and axis lable arguments, the following patterns are substituted:
"%C"
class of argument object
"%A"
deparsed argument object
"%D"
time/date-string when the plot was generated
If argument ...
contains argument ylim
, this may either be
as in plot.default
(i.e. a vector of length 2) or a vector of
length 2*(number of plotted dimensions + e), where e is 1 or 0 depending
on whether absolute information is plotted or not;
in the case of longer length,
if e is 1, the first two elements are the values for ylim
in panel "Abs", while the last 2*(number of plotted dimensions)
are the values for ylim
for the plotted dimensions of the IC,
one pair for each dimension.
Similarly, if argument ...
contains arguments xaxt
or
yaxt
, these may be vectorized, with one value for each of the panels
to be plotted. This is useful for stacking panels over each other, using
a common x-axis (see example below).
The ...
argument may also contain an argument withbox
which
if TRUE
warrants that even if xaxt
and yaxt
both are
FALSE
, a box is drawn around the respective panel.
In addition, argument ...
may contain arguments panel.first
,
panel.last
, i.e., hook expressions to be evaluated at the very beginning
and at the very end of each panel (within the then valid coordinates).
To be able to use these hooks for each panel individually, they may also be
lists of expressions (of the same length as the number of panels and
run through in the same order as the panels).
Kohl, M. (2005) Numerical Contributions to the Asymptotic Theory of Robustness. Bayreuth: Dissertation.
L2ParamFamily-class
, IC-class
N <- NormLocationScaleFamily(mean=0, sd=1)
IC1 <- optIC(model = N, risk = asCov())
infoPlot(IC1)
## don't run to reduce check time on CRAN
# \donttest{
## selection of subpanels for plotting
par(mfrow=c(1,2))
infoPlot(IC1, mfColRow = FALSE, to.draw.arg=c("Abs","sd"))
infoPlot(IC1, mfColRow = FALSE, to.draw.arg=c("Abs","sd"), log="y")
infoPlot(IC1, mfColRow = FALSE, to.draw.arg=c("Abs","mean"),
panel.first= grid(), ylim = c(0,4), xlim = c(-6,6))
infoPlot(IC1, mfColRow = FALSE, to.draw.arg=c("Abs","mean"),
panel.first= grid(), ylim = c(0,4,-3,3), xlim = c(-6,6))
par(mfrow=c(1,3))
infoPlot(IC1, mfColRow = FALSE, panel.first= grid(),
ylim = c(0,4,0,.3,0,.8), xlim=c(-6,6))
par(mfrow=c(1,1))
data <- r(N)(20)
par(mfrow=c(1,3))
infoPlot(IC1, data=data, mfColRow = FALSE, panel.first= grid(),
with.lab = TRUE, cex.pts=2,
which.lbs = c(1:4,15:20), which.Order = 1:6,
return.Order = TRUE)
infoPlot(IC1, data=data[1:10], mfColRow = FALSE, panel.first= grid(),
with.lab = TRUE, cex.pts=0.7)
par(mfrow=c(1,1))
ICr <- makeIC(list(function(x)sign(x),function(x)sign(abs(x)-qnorm(.75))),N)
data <- r(N)(600)
data.c <- c(data, 1000*data[1:30])
par(mfrow=c(3,1))
infoPlot(ICr, data=data.c, tmar=c(4.1,0,0), bmar=c(0,0,4.1),
xaxt=c("n","n","s"), mfColRow = FALSE, panel.first= grid(),
cex.pts=c(.9,.9), alpha.trsp=20, lwd=2, lwdI=1.5, col=3,
col.pts=c(3,2), colI=2, pch.pts=c(20,20), inner=FALSE,
scaleX = TRUE, scaleX.fct=pnorm, scaleX.inv=qnorm,
scaleY=TRUE, scaleY.fct=function(x) pchisq(x,df=1),
scaleY.inv=function(x)qchisq(x,df=1),legend.cex = 1.0)
# }
Run the code above in your browser using DataLab