dqf.outlier: DQF computation for data frame

In GabeChandler/dqfAnomaly: Visual Anomaly Detection via Depth Quantile Functions

Description

Computes the averaged DQFs for a given data set

Usage

dqf.outlier(
  data = NULL,
  gram.mat = NULL,
  g.scale = 2,
  angle = c(30, 45, 60),
  kernel = "linear",
  p1 = 1,
  p2 = 0,
  n.splits = 100,
  subsample = 50,
  z.scale = TRUE,
  k.w = 3,
  adaptive = TRUE,
  G = "norm"
)

Arguments

data	data frame or matrix
gram.mat	alternatively, the precompute Gram matrix
g.scale	scales the base distribution G
angle	3-vector of angles (must be between 0 and 90)
kernel	of form "linear", "rbf" or "poly", or a user defined function
p1	first parameter for kernel (if required)
p2	second parameter for kernel (if required)
n.splits	integer number of base points to compute DQF at
subsample	integer number of random pairs to base averaged DQF on
z.scale	logical to z-scale data prior to analysis
k.w	integer number of points to windsorize at each extreme of data
adaptive	logical to use adaptive DQF (windsorized st dev)
G	base distribution: "norm" or "unif"
angles	of cone from midline, must live between 0 and 90

Examples

fit.dqf <- dqf.outlier(iris[51:102,1:4], g.scale=6)

GabeChandler/dqfAnomaly documentation built on Jan. 25, 2022, 12:33 a.m.