confidence_plot: Plot to check reported change-points

In blockcpd: Change Point Detection for Multiple Aligned Independent Time Series

Plot to check reported change-points

Description

Plots the estimates of how likely it is for the model to detect a change at any given point. True change-points should have confidence near $100%$, while non change-points should have a confidence near $0%$. It might also be difficult to detect a true change-point at the given sample size. In this case, it should fluctuate in the middle.

Usage

confidence_plot(
  model,
  scale = "percentage",
  index_values = NULL,
  index_variable_name = "Index",
  pkg = "base"
)

Arguments

model	A blockcpd model object.
scale	A string describing the scale which the y-scale should is plotted. Possible values are "percentage", "probability" and "frequency".
index_values	A numerical vector of size ncol that contains the values of the the variable corresponding to the change points.
index_variable_name	Name of the variable segmented.
pkg	Graphical package to be used for plotting. Current values are "base".

Value

No return value.

Examples

td = rcpd(nrow = 10, ncol = 10)
model = fit_blockcpd(td$data_matrix, bootstrap = TRUE)
confidence_plot(model)

blockcpd documentation built on Aug. 12, 2022, 5:07 p.m.