princurve

Fitting a principal curve to a data matrix in arbitrary dimensions. A principal curve is a smooth curve passing through the middle of a multidimensional dataset. This package is an R/C++ reimplementation of the S/Fortran code provided by Trevor Hastie, with multiple performance tweaks.

Example

Usage of princurve is demonstrated with a toy dataset.

t <- runif(100, -1, 1)
x <- cbind(t, t ^ 2) + rnorm(200, sd = 0.05)
colnames(x) <- c("dim1", "dim2")

plot(x)

A principal curve can be fit to the data as follows:

library(princurve)
fit <- principal_curve(x)
plot(fit); whiskers(x, fit$s, col = "gray")

See ?principal_curve for more information on how to use the princurve package.

Latest changes

Check out news(package = "princurve") for a full list of changes.

Recent changes in princurve 2.1.4 (2019-05-29)

• Fix warning in stats::approx() due to changes made in R 3.6.

• Defuncted principal.curve() and get.lam().

Recent changes in princurve 2.1.3 (2018-09-10)

• Removed extra dependencies in princurve by removing vignettes; fixes #28.

• Fully deprecated principal.curve() and get.lam().

References

Hastie, T. and Stuetzle, W., Principal Curves, JASA, Vol. 84, No. 406 (Jun., 1989), pp. 502-516, DOI: 10.2307/2289936 (PDF)