R: Get best candidate change point

anotherBkp

R Documentation

Get best candidate change point

Description

Get best candidate change point according to binary segmentation

Usage

anotherBkp(Y, weightFUN = defaultWeights, verbose = FALSE)

Arguments

`Y`	A `n*p` matrix, `p` signals of length `n` to be segmented (centered by column)
`weightFUN`	A `function` returning a `(n-1)*1` vector of weights for the candidate change point positions. Default weights yield the likelihood ratio test (LRT) statistic for the identification of a single change point.
`verbose`	A `logical` value: should extra information be output ? Defaults to `FALSE`.

Details

Contrary to oneBkp, anotherBkp handles missing values (NA:s).

Value

A numerical value, a candidate breakpoint position

Author(s)

Morgane Pierre-Jean and Pierre Neuvial

Examples


p <- 2
n <- 100

sim <- randomProfile(n, 1, 1, p)
Y <- sim$profile
bkp <- anotherBkp(Y)
print(bkp)
print(oneBkp(Y))
##  stopifnot(identical(oneBkp(Y), bkp))
plotSeg(Y, list(sim$bkp, bkp))

## robustness to NA:s
h <- 2
idxs <- seq(from=max(sim$bkp[1]-h, 1), min(sim$bkp[1]+h, n))
Y[idxs, p] <- NA
oneBkp(Y)  ## does not work
bkp <- anotherBkp(Y)  ## works
bkp-sim$bkp