fdars (version 0.3.3)

depth: Depth Functions for Functional Data

Description

Functions for computing various depth measures for functional data. Compute Functional Data Depth

Usage

depth(
  fdataobj,
  fdataori = NULL,
  method = c("FM", "mode", "RP", "RT", "BD", "MBD", "MEI", "FSD", "KFSD", "RPD"),
  ...
)

Value

A numeric vector of depth values, one per curve in fdataobj.

Arguments

fdataobj

An object of class 'fdata' to compute depth for.

fdataori

An object of class 'fdata' as reference sample. If NULL, uses fdataobj as reference.

method

Depth method to use. One of "FM" (Fraiman-Muniz), "mode" (modal), "RP" (random projection), "RT" (random Tukey), "BD" (band depth), "MBD" (modified band depth), "MEI" (modified epigraph index), "FSD" (functional spatial depth), "KFSD" (kernel functional spatial depth), or "RPD" (random projection with derivatives). Default is "FM".

...

Additional arguments passed to the specific depth function.

Details

Unified interface for computing various depth measures for functional data.

Available methods:

FM

Fraiman-Muniz depth - integrates univariate depths over domain

mode

Modal depth - based on kernel density estimation

RP

Random projection depth - projects to random directions

RT

Random Tukey depth - halfspace depth via random projections

BD

Band depth - proportion of bands containing the curve (1D only)

MBD

Modified band depth - allows partial containment (1D only)

MEI

Modified epigraph index - proportion of time below other curves (1D only)

FSD

Functional spatial depth - based on spatial signs

KFSD

Kernel functional spatial depth - smoothed FSD

RPD

Random projection with derivatives - includes curve derivatives

Examples

fd <- fdata(matrix(rnorm(100), 10, 10))

# Different depth methods
depth(fd, method = "FM")
depth(fd, method = "mode")
depth(fd, method = "RP")