CPOP wrapper — optimal change-in-slope detection — cpop

Wraps cpop::cpop() (Fearnhead, Maidstone and Letchford, 2019; Fearnhead and Grose, 2024): exact penalised estimation of a continuous piecewise-linear mean via dynamic programming with functional pruning. This is the engine behind cpt_detect(change_in = "slope"). The fitted broken line is stored in the fitted column and renders via autoplot(show_fit = TRUE).

Usage

cpop_wrapper(x, penalty = NULL, sd = NULL, ...)

Arguments

x: A numeric vector.
penalty: Penalty for adding a changepoint. Defaults to 2 * log(length(x)).
sd: Noise standard deviation; when NULL it is estimated from the data by the engine's default difference-based estimator.
...: Additional arguments passed to cpop::cpop().

Value

A ggcpt object with change_in = "slope". For a continuous fit the reported location is the kink point itself.

References

Fearnhead P, Maidstone R, Letchford A (2019). “Detecting changes in slope with an L0 penalty.” Journal of Computational and Graphical Statistics, 28(2), 265–275.

Fearnhead P, Grose D (2024). “cpop: Detecting changes in piecewise-linear signals.” Journal of Statistical Software, 109(7), 1–30.

Examples

set.seed(2026)
y <- cumsum(c(rep(0.3, 100), rep(-0.4, 100))) + rnorm(200)
res <- cpop_wrapper(y)
res$changepoints
#> # A tibble: 1 × 2
#>      cp cp_value
#>   <int>    <dbl>
#> 1   100     30.4
ggplot2::autoplot(res, show_fit = TRUE)