BUHDA

Bottom-Up High-Dimensional Adaptive change-point detection.

Description

This repository contains implementation of the BUHDA algorithm described in Maeng, Wang and Fryzlewicz (2026). BUHDA is designed to detect multiple change-points for high-dimensional time series where change in mean is of interest.

Installation

# install.packages("remotes")
remotes::install_github("hmaeng/BUHDA")

Usage

library(BUHDA)

# simulate some data
set.seed(1)
n <- 100
p <- 10
k <- c(1, 10)
eta <- c(30, 70)
varthetas <- c(3, 4)
x <- change_data_generator(n=n, p=p, k=k, eta=eta, varthetas=varthetas)$x

# run the main function
p <- dim(x)[1]
n <- dim(x)[2]
thr.2 <- GETthreshold(n, p, method="agg2", M=n, qntl=.95, nrep=100, show_progress=TRUE)
thr.inf <- GETthreshold(n, p, method="aggoo", M=n, qntl=.95, nrep=100, show_progress=TRUE)
res <- BU(x=x, thr.2=thr.2, thr.inf=thr.inf, sparsity="unknown")
res$cpt

Reference

If you use this package, please cite:

Maeng, H., Wang, T., and Fryzlewicz, P. (2026). High-dimensional sparsity-adaptive multiple change-point detection. arXiv preprint arXiv:2607.20928