cpt_batch: Batch changepoint detection over many series

In ggchangepoint: Combines Changepoint Analysis with 'ggplot2'

Batch changepoint detection over many series

Description

Runs one detector over every series in a collection: the panel-data loop that methodological and applied work both need constantly. Accepts a matrix/data frame (one column per series) or a named list of numeric vectors. Honours future::plan() for parallel execution when the future.apply package is available, with parallel-safe RNG.

Usage

cpt_batch(
  x,
  method = "pelt",
  change_in = "mean",
  index = NULL,
  seed = NULL,
  keep_fit = TRUE,
  ...
)

## S3 method for class 'ggcpt_batch'
print(x, ...)

## S3 method for class 'ggcpt_batch'
tidy(x, ...)

## S3 method for class 'ggcpt_batch'
autoplot(object, ...)

Arguments

x	For cpt_batch(), a numeric matrix or data frame (columns are series) or a list of numeric vectors; for the print() and tidy() methods, a ggcpt_batch object.
method	Detection method, passed to cpt_detect().
change_in	What to detect change in, passed to cpt_detect().
index	Optional time index shared by every series in the panel (a vector of dates, say), or a named list of one index per series. A list has to cover every series, with indices of one type, because tidy() and autoplot() stack them into one table and one axis. Carried onto each result and used by tidy() and autoplot(), so a faceted plot of fifty series shows dates rather than positions.
seed	Optional seed for reproducible parallel execution (passed to future.apply::future_lapply() as future.seed; applied via set.seed() when running sequentially). The seed is scoped to this call: .Random.seed is saved and restored, so a seeded call inside a simulation loop does not pin the loop's own stream.
keep_fit	Keep each engine's raw fit in result[[i]]$fit? Defaults to TRUE, which is what makes a batch result as inspectable as a single one. Set it to FALSE for a large panel: a few engines return fits far bigger than the data they were given. strucchange keeps a triangular O(n^2) RSS matrix, so a single 2000-point series costs about 135 MB, and bfast and bocpd are in the tens of MB, and a panel multiplies that by the number of series. Everything else on the object, including tidy() and autoplot(), is unaffected; only accessors that read $fit (cpt_statistic(), ggcpt_posterior(), cpt_confint(engine = \"native\")) need it.
...	Additional arguments passed to every cpt_detect() call.
object	A ggcpt_batch object (for autoplot()).

Value

A ggcpt_batch object: a tibble with one row per series and columns series, n_changepoints, changepoints (a list-column of tidy tibbles), and result (a list-column of ggcpt objects). Methods: print(), tidy() (one row per changepoint across all series, with columns series, cp and cp_value), and autoplot() (faceted small-multiples with each series' changepoints).

Examples

set.seed(2026)
X <- cbind(a = c(rnorm(60), rnorm(60, 4)), b = rnorm(120))
batch <- cpt_batch(X, method = "pelt")
batch
tidy(batch)
ggplot2::autoplot(batch)

ggchangepoint documentation built on Nov. 5, 2026, 9:06 a.m.