Methods for data segmentation under a sparse regression model. See
High-dimensional data segmentation in regression settings permitting heavy tails and temporal dependence, Haeran Cho and Dom Owens, arxiv.org/abs/2209.08892
package moseg installable via
devtools::install_github("https://github.com/Dom-Owens-UoB/moseg")
We can simulate from a piecewise sparse regression model via
set.seed(111)
dat <- moseg.sim(500, 50, q = 2, kappa = 4)
Identify change points:
out <- moseg(dat$X, dat$y, 100, do.scale = FALSE)
Multiscale:
out <- moseg.ms(dat$X, dat$y, c(50,100,150), do.scale = FALSE)
Using cross-validation:
cv <- moseg.cv(dat$X, dat$y, 50, do.scale = FALSE)
cv.ms <- moseg.ms.cv(dat$X, dat$y, c(50,100,150), do.scale = FALSE)