cpss.em {cpss}

Detecting changes in exponential family

Description

Detecting changes in exponential family

Usage

cpss.em(
  dataset,
  family,
  size = NULL,
  algorithm = "BS",
  dist_min = floor(log(n)),
  ncps_max = ceiling(n^0.4),
  pelt_pen_val = NULL,
  pelt_K = 0,
  wbs_nintervals = 500,
  criterion = "CV",
  times = 2
)

Arguments

dataset

a numeric matrix of dimension n\times d, where each row represents an observation and each column stands for a variable. A numeric vector is also acceptable for univariate observations.

family

a character string specifying the underlying distribution. In the current version, detecting changes in binomial ("binom"), multinomial ("multinom"), Poisson ("pois"), exponential ("exp"), geometric ("geom"), Dirichlet ("diri"), gamma ("gamma"), beta ("beta"), chi-square ("chisq") and inverse gaussian ("invgauss") distributions are supported.

size

an integer indicating the number of trials only if family = "binom" or family = "multinom".

algorithm

a character string specifying the change-point searching algorithm, one of the following choices: "SN" (segment neighborhood), "BS" (binary segmentation), "WBS" (wild binary segmentation) and "PELT" (pruned exact linear time) algorithms.

dist_min

an integer specifying minimum searching distance (length of feasible segments).

ncps_max

an integer specifying an upper bound of the number of true change-points.

pelt_pen_val

a numeric vector specifying candidate values of the penalty only if algorithm = "PELT".

pelt_K

a numeric value for pruning adjustment only if algorithm = "PELT". It is usually taken to be 0 if the negative log-likelihood is used as a cost, see Killick et al. (2012).

wbs_nintervals

an integer specifying the number of random intervals drawn only if algorithm = "WBS", see Fryzlewicz (2014).

criterion

a character string specifying the model selection criterion, "CV" ("cross-validation") or "MS" ("multiple-splitting").

times

an integer specifying how many times of sample-splitting should be performed; It should be 2 if criterion = "CV".

Value

cpss.em returns an object of an S4 class, called "cpss", which collects data and information required for further change-point analyses and summaries. See cpss.custom.

References

Killick, R., Fearnhead, P., and Eckley, I. A. (2012). Optimal Detection of Changepoints With a Linear Computational Cost. Journal of the American Statistical Association, 107(500):1590–1598.

Fryzlewicz, P. (2014). Wild binary segmentation for multiple change-point detection. The Annals of Statistics, 42(6): 2243–2281.

See Also

cpss.meanvar cpss.mean cpss.var

Examples

library("cpss")
set.seed(666)
n <- 1000
tau <- c(100, 300, 700, 900)
tau_ext <- c(0, tau, n)
theta <- c(1, 0.2, 1, 0.2, 1)
seg_len <- diff(c(0, tau, n))
y <- unlist(lapply(seq(1, length(tau) + 1), function(k) {
  rexp(seg_len[k], theta[k])
}))
res <- cpss.em(
  y, family = "exp", algorithm = "WBS", ncps_max = 10,
  criterion = "MS", times = 10
)
cps(res)
# [1] 100 299 705 901