detect.cps: Multiple change point detection in the network (or...

In fabisearch: Change Point Detection in High-Dimensional Time Series Networks

Multiple change point detection in the network (or clustering) structure of multivariate high-dimensional time series

Description

This function detects multiple change points in the network (or clustering) structure of multivariate high-dimensional time series using non-negative matrix factorization and a binary search.

Usage

detect.cps(
  Y,
  mindist = 35,
  nruns = 50,
  nreps = 100,
  alpha = NULL,
  rank = NULL,
  algtype = "brunet",
  testtype = "t-test",
  ncore = 1
)

Arguments

Y	An input multivariate time series in matrix format, with variables organized in columns and time points in rows. All entries in Y must be positive.
mindist	A positive integer with default value equal to 35. It is used to define the minimum distance acceptable between detected change points.
nruns	A positive integer with default value equal to 50. It is used to define the number of runs in the NMF function.
nreps	A positive integer with default value equal to 100. It is used to define the number of permutations for the statistical inference procedure.
alpha	A positive real number with default value set to NULL. When alpha = NULL, then the p-value calculated for inference on the change points is returned. If alpha = a positive integer value, say 0.05, then it is used to define the significance level for inference on the change points.
rank	A positive integer, which defines the rank used in the optimization procedure to detect the change points. If rank = NULL, which is also the default value, then the optimal rank is computed. If rank = a positive integer value, say 4, then a predetermined rank is used.
algtype	A character string, which defines the algorithm to be used in the NMF function. By default it is set to "brunet". See the "Algorithms" section of nmf for more information on the available algorithms.
testtype	A character string, which defines the type of statistical test to use during the inference procedure. By default it is set to "t-test". The other options are "ks" and "wilcox" which correspond to the Kolmogorov-Smirnov and Wilcoxon tests, respectively.
ncore	A positive integer with default value equal to 1. It is used to define the number of cores to use in the procedure

Value

A list with the following components :
rank: The rank used in the optimization procedure for change point detection.
change_points: A table of the detected change points where column "T" is the time of the change point and "stat_test" is the result (either a boolean value if alpha = a positive real number, or the p-value if alpha = NULL) of the t-test.
compute_time: The computational time, saved as a "difftime" object.

Author(s)

Martin Ondrus, [e-mail], Ivor Cribben, [e-mail]

References

"Factorized Binary Search: a novel technique for change point detection in multivariate high-dimensional time series networks", Ondrus et al. (2021), <arXiv:2103.06347>.

Examples

## Change point detection for a multivariate data set, sim2, using settings:
## rank = 3, mindist = 99, nruns = 2, and nreps = 2
set.seed(123)
detect.cps(sim2, rank = 3, mindist = 99, nruns = 2, nreps = 2)


# $rank
# [1] 3
#
# $change_points
#     T stat_test
# 1 101 0.3867274
#
# $compute_time
# Time difference of 0.741534 mins

fabisearch documentation built on Jan. 12, 2023, 5:08 p.m.