Bayesian online changepoint detection wrapper (BOCPD)

Wraps ocp::onlineCPD(), an implementation of Bayesian Online Changepoint Detection (Adams and MacKay, 2007). BOCPD recursively updates a posterior over the current run length (time since the last change); the maximum a posteriori set of changepoints is reported, and the full run-length posterior is kept so that ggcpt_runlength() can draw the signature run-length heatmap.

Usage

bocpd_wrapper(x, hazard = 100, ...)

Arguments

x

A numeric vector.

hazard

Constant hazard rate \(1/\lambda\) of the change process; larger hazard values mean changes are expected less often. Defaults to 100 (the upstream default).

...

Additional arguments passed to ocp::onlineCPD().

Value

A ggcpt object with the MAP changepoint set. The full ocp fit (including the run-length posterior) is kept in $fit.

References

Adams RP, MacKay DJ (2007). “Bayesian online changepoint detection.” arXiv preprint arXiv:0710.3742.

Examples

res <- bocpd_wrapper(c(rnorm(60), rnorm(60, 4)))
res$changepoints
#> # A tibble: 1 × 2
#>      cp cp_value
#>   <int>    <dbl>
#> 1    60     1.10
ggcpt_runlength(res)