SemimetricDeriv: SemimetricDeriv

In sipemu/Nonparametric-Functional-Data-Analysis: Nonparametric Functional Data Analysis

Description

SemimetricDeriv is a function that calculates a distance matrix based on the L_2-norm of the derivatives of the functional data. A Rcpp port of the semimetric function based on derivatives of Ferraty and Vieu.

Usage

SemimetricDeriv(Data1, 
                Data2, 
                q = 2, nknot = 20, range.grid = c(0,1), 
                Hhalf = NULL)

Arguments

Data1	Matrix with functional data (curves) each row one; n rows
Data2	Matrix with a second set of functional data (curves) each row one; m rows
q	Order of derivation
nknot	Number of the interior knots for the B-spline basis
range.grid	Vector of length two, containing the domain of the curves
Hhalf	A design matrix that is based on the B-spline basis design

Value

Semimetric returns a list with: semimetric as the n x m - distance-matrix Hhalf A design matrix that is based on the B-spline basis design, which can be reused in following semimetric calculations

Author(s)

Simon Mueller [e-mail]

References

http://www.math.univ-toulouse.fr/staph/npfda/

Ferraty, F. and Vieu, P. Nonparametric Functional Data Analysis. Springer 2006.

See Also

Semimetric

Examples

# functional data sets
require (fds)

# fat spectrum dataset
X <- t(Fatspectrum$y)

# setup semimetric parameters
q <- 2
nknot <- 20
range.grid <- c (min (Fatspectrum$x), 
                 max (Fatspectrum$x))

# calculate distance matrix
Dist1 <- SemimetricDeriv (X[1:100, ], 
                          X[101:150, ], 
                          q, nknot, range.grid)

# calculate distance matrix using Hhalf
Dist2 <- SemimetricDeriv (X[1:100, ], 
                          X[151:215, ], 
                          q, nknot, range.grid, 
                          Dist1$Hhalf)

sipemu/Nonparametric-Functional-Data-Analysis documentation built on May 29, 2019, 10:10 p.m.