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"),
...
)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")
#> [1] 0.48 0.46 0.40 0.66 0.52 0.36 0.52 0.66 0.52 0.42
depth(fd, method = "mode")
#> [1] 0.2714079 0.1717654 0.2102241 0.2917229 0.2032817 0.1923078 0.2250097
#> [8] 0.2894317 0.2493039 0.1962121
depth(fd, method = "RP")
#> [1] 0.3072727 0.2254545 0.2581818 0.2854545 0.2509091 0.2109091 0.2818182
#> [8] 0.3181818 0.2981818 0.2909091