sde (version 2.0.18)

MOdist: Markov Operator distance for clustering diffusion processes.

Description

Markov Operator distance for clustering diffusion processes.

Usage

MOdist(x, M=50, rangeval=range(x, na.rm=TRUE, finite = TRUE))

Value

X

a dist object

Arguments

x

one or multi-dimensional time series.

M

number of splines bases used to approximate the Markov Operator.

rangeval

a vector containing lower and upper limit. Default is the range of x.

Author

Stefano Maria Iacus

Details

This function return a lower triangular dist object to be further used in cluster analysis (see examples below).

If x is a one-dimensional time series, the output is the scalar 0, not a dist object.

If x has less than 2 observations, NA is returned.

If time series x contains missing data, then x is converted to a zoo object and missing data are imputed by interpolation.

References

De Gregorio, A. Iacus, S.M. (2008) Clustering of discretely observed diffusion processes, Computational Statistics and Data Analysis, 54(12), 598-606, tools:::Rd_expr_doi("10.1016/j.csda.2009.10.005").

Examples

if (FALSE) {
data(quotes)

plot(quotes)

d <- MOdist(quotes)
cl <- hclust( d )
groups <- cutree(cl, k=4)

cmd <- cmdscale(d)
plot( cmd, col=groups)
text( cmd, labels(d) , col=groups)

plot(quotes, col=groups)

plot(quotes, col=groups,ylim=range(quotes))
}