SLIP.thresh.d: SLIP with thresholding (d version)

In MengtaoWen/SLIP: Change Testing for Large-Scale Datastreams with FDR Control

SLIP with thresholding (d version)

Description

Use SLIP with mean screening to detect abnormal data streams each of which occurs at least one change.

Usage

SLIP.thresh.d(
  dat,
  alpha,
  upperPi = 0.5,
  r = 3,
  covEst = T,
  estMthd = "Cholesky",
  trueCov = NULL,
  outputW = FALSE,
  outputCP = FALSE
)

Arguments

dat	n x p matrix (p features, n observations)
alpha	FDR nominal level
upperPi	retained proportion after thresholding 0 < d < 1(default 0.5)
r	splitting ratio, (r-1) pieces versus 1 piece
covEst	Estimate covariance or not (logical); T for Est
estMthd	optional estimation methods c("Cholesky", "POET")
trueCov	the true covariance matrix; only optional when covEst=F
outputW	a logical parameter FALSE(default); if TRUE, the W-statistics and the threshold will be returned.
outputCP	logical parameter FALSE(default); if TRUE, the change-point location in (0, 1) corresponding to signals will be returned.

Value

A list contains:

sig	indices of signals
FDP	estiamted FDP
W	W-statistic, optional only when W = TRUE
L	threshold, optional only when W = TRUE
cps	change-points, optional only when outputCP = TRUE

Examples

  N = 120; p = 200
  data = SLIP.scp.generator(N, p)
  SLIP.thresh.d(data$dat, 0.1)

MengtaoWen/SLIP documentation built on May 3, 2022, 6:45 a.m.