Takes an object of class lpc, lpc.spline or ms. In the case of principal curves, it plots any subset of the following components of the local principal curve: Centers of mass; the curve connecting the local centers of mass; the cubic spline representation of the curve; the projections onto the curve; the starting points. For the mean shift procedure, it produces a plot of mean shift trajectories and cluster centers.
# S3 method for lpc
plot(x, type, unscale = TRUE, lwd = 1, datcol = "grey60",
datpch = 21, masscol = NULL, masspch = 15, curvecol = 1, splinecol = 3,
projectcol = 4, startcol = NULL, startpch=NULL,...)
# S3 method for lpc.spline
plot(x, type, unscale = TRUE, lwd = 1, datcol = "grey60",
datpch = 21, masscol = NULL, masspch = 15, curvecol = 1, splinecol = 3,
projectcol = 4, startcol = NULL, startpch=NULL,...)
# S3 method for ms
plot(x, unscale=FALSE, lwd=1, datcol="grey70", datpch=21, masscol=NULL,
masspch=15, curvecol=NULL, ...)A plot of adequate dimensionality (depending on the type of object).
For local principal curves, the minimum supported dimension is \(d=2\), and for the mean shift it is \(d=1\). In either case, the maximum supported dimension is \(d=16\). With increasing dimension \(d\), less plotting options tend to be supported. The nicest plots are obtained for \(d=2\) and \(d=3\).
The most flexible plotting option is masscol. Depending on the
length of the specified vector, this will be interpreted differently. If
a scalar is provided, the corresponding color will be given to all centers of
mass (or cluster centers). For LPCs, if the length of the vector is larger than 1, then this option
will assign different colours to different depths, or different branch
numbers, or to individual data points, depending on the length. The
default setting is assigning colours according to depth, in the order
red, blue, black.
an object of class lpc, lpc.spline, or ms.
a vector of type c("mass", "spline",...) with possible entries mass, curve, spline, project, start.
if TRUE, then data (and all fitted components) are scaled back to their original scale; otherwise the scaled data are plotted (only relevant if scaled=TRUE in the fitted object). For ms, this is currently unimplemented.
width of principal curves or trajectories.
color of data points.
plotting symbol for data points.
color of centers of mass (see below) or cluster centers.
plotting symbol for centers of mass or cluster centers.
color of the curve interpolating the local centers of mass (this is the "local principal curve"!).
color of the spline representation of the local principal curve.
color of projections onto the spline representation of the local principal curve.
color of the plotted starting points.
plotting symbol for starting points; needs to be either a single symbol, or a vector of symbols of the same length as the number of starting points.
further arguments passed to plot or scatterplot3d.
This function computes all missing information (if possible), so computation will take the longer the less informative the given object is, and the more advanced aspects are asked to plot!
JE
Ameijeiras-Alonso, J. and Einbeck, J. (2023). A fresh look at mean-shift based modal clustering, Advances in Data Analysis and Classification, tools:::Rd_expr_doi("10.1007/s11634-023-00575-1").
Einbeck, J., Tutz, G., and Evers, L. (2005). Local principal curves. Statistics and Computing 15, 301-313.
Einbeck, J., Evers, L. & Hinchliff, K. (2010): Data compression and regression based on local principal curves. In A. Fink, B. Lausen, W. Seidel, and A. Ultsch (Eds), Advances in Data Analysis, Data Handling, and Business Intelligence, Heidelberg, pp. 701--712, Springer.
lpc, lpc.spline , ms
data(calspeedflow)
lpc1 <- lpc(calspeedflow[,3:4])
plot(lpc1, type=c("spline","project"), lwd=2)
ms1<- ms(calspeedflow[,3:4], subset=sample.int(444,100), plot=FALSE)
# starts trajectories from 100 random obs'n
plot(ms1, masscol=1)
plot(ms1, curvecol="grey30")
data(mussels, package="dr")
ms2 <- ms(mussels[,-3], scaled=1, h=0.1, plot=FALSE)
plot(ms2, datpch=20, masspch=24)
Run the code above in your browser using DataLab