FuNopaCl: Nonparametric Classification for Functional Data

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

Description

FuNopaCl is a function that estimates optimal bandwidth by k nearest neighbour local cross-validation for the kernel estimate based on the learning data.

Usage

FuNopaCl(X, classes, semimetric, semimetric.params)

Arguments

X	Matrix with the functional data (curves) each row one
classes	Vector of the classes
semimetric	A string of choosing the semimetric; allowed are: "Deriv" and "PCA"
semimetric.params	Parameters for the semimetric function.

Value

FuNopaCl returns an object of the class FuNopaCl;

Author(s)

Simon Mueller [e-mail]

References

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

See Also

Semimetric, predict.FuNopaCl

Examples

# functional datasets
require (fds)

# fat spectrum dataset
Y <- Fatvalues
X <- t(Fatspectrum$y)
Y[Y < 20] <- 1
Y[Y >= 20] <- 2

# setup semimetric params
semimetric.params <- c()
semimetric.params$q <- 2
semimetric.params$nknot  <- 20
semimetric.params$range.grid  <- c (min (Fatspectrum$x), 
                                    max (Fatspectrum$x))
# learn and testsample
learn <- 1:160
test <- 161:215

# parameter estimation and prediction by cross-validation
Learn.Fat.CV <- FuNopaCl (X[learn, ], 
                          Y[learn], 
                          semimetric = "Deriv", 
                          semimetric.params)

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