Robust multiscale time-average variance estimation for change point detection.
Software accompanying
E. T. McGonigle and H. Cho (2023) "Robust multiscale estimation of time-average variance for time series segmentation".
- The main routines are contained in main.R.
To perform robust TAVC estimation, do the following:
- Source main.R into
R
. - Read the description for
robust.tavc.est
within main.R.
To perform mean change point detection with the robust TAVC estimation procedure
- Using the multiscale bottom-up MOSUM procedure: install the MOSUM R package and read the description in `mosum.tavc'.
- Using the wild binary segmentation 2 algorithm: install the breakfast R package and read the description in `WBS2.tavc'.
For example,
cpt.sig = c(rep(0,200),rep(2,300),rep(4,200),rep(2,300))
set.seed(123)
x = cpt.sig + arima.sim(model = list(ar = 0.5), sd = sqrt(1-0.5^2), n = 1000)
x.m.c = mosum.tavc(x,G = c(30,60,90,150), alpha = 0.05)
x.m.c$cpts
x.w.c = wbs2.tavc(x, min.int.len = 60)
x.w.c$cpts
If you have any questions, please contact euan.mcgonigle@bristol.ac.uk