Fitting multivariate data patterns with local principal curves;
including simple tools for data compression (projection), bandwidth
selection, and measuring goodness-of-fit.
This package implements the techniques introduced in Einbeck, Tutz
& Evers (2005), and successive related papers.
The main functions to be called by the user are
lpc, for the estimation of the local centers of mass
which make up the principal curve;
lpc.spline, which is a smooth and fully parametrized
cubic spline respresentation of the latter;
lpc.project, which enables to compress data by
projecting them orthogonally onto the curve;
lpc.coverage and Rc for assessing
goodness-of-fit;
lpc.self.coverage for bandwidth selection;
the generic plot and print methods for objects
of class lpc and lpc.spline.
This package also contains some code for density
mode detection (‘local principal points’) and mean shift clustering (as well as bandwidth
selection in this context), which implements the methods presented in
Einbeck (2011). See the help file for ms.
A second R package which will implement the extension of local principal
curves to local principal surfaces and manifolds, as proposed in
Einbeck, Evers & Powell (2010), is in preparation.
Details
Package:
LPCM
Type:
Package
License:
GPL (>=2)
LazyLoad:
yes
Acknowledgements
Contributions (in form of pieces of code, or useful suggestions for
improvements) by Jo Dwyer, Mohammad Zayed, and
Ben Oakley are gratefully acknowledged.
Einbeck, J., Tutz, G., & Evers, L. (2005): Local principal curves, Statistics and Computing 15, 301-313.
Einbeck, J., Evers, L., & Powell, B. (2010): Data compression and regression through local principal curves and surfaces, International Journal of Neural Systems,
20, 177-192.
Einbeck, J. (2011): Bandwidth selection for nonparametric unsupervised
learning techniques – a unified approach via self-coverage. Journal of
Pattern Recognition Research 6, 175-192.