fastcpd: Fast Change Point Detection via Sequential Gradient Descent

Implements fast change point detection algorithm based on the paper "Sequential Gradient Descent and Quasi-Newton's Method for Change-Point Analysis" by Xianyang Zhang, Trisha Dawn <>. The algorithm is based on dynamic programming with pruning and sequential gradient descent. It is able to detect change points a magnitude faster than the vanilla Pruned Exact Linear Time(PELT). The package includes examples of linear regression, logistic regression, Poisson regression, penalized linear regression data, and whole lot more examples with custom cost function in case the user wants to use their own cost function.

Version: 0.14.3
Depends: R (≥ 2.10)
Imports: fastglm, forecast, glmnet, Matrix, methods, Rcpp (≥ 0.11.0), stats, tseries
LinkingTo: progress, Rcpp, RcppArmadillo, RcppClock, testthat
Suggests: abind, breakfast, changepoint, cpm, CptNonPar, dplyr, fpop, ggplot2, gridExtra, jointseg, knitr, lubridate, matrixStats, mockthat, mvtnorm, not, numDeriv, RcppClock, reshape2, rmarkdown, segmented, stepR, testthat (≥ 3.0.0), VARDetect, wbs, xml2, zoo
Published: 2024-04-26
DOI: 10.32614/CRAN.package.fastcpd
Author: Xingchi Li ORCID iD [aut, cre, cph], Xianyang Zhang [aut, cph]
Maintainer: Xingchi Li < at>
License: GPL (≥ 3)
NeedsCompilation: yes
Citation: fastcpd citation info
Materials: README NEWS
In views: TimeSeries
CRAN checks: fastcpd results


Reference manual: fastcpd.pdf
Vignettes: Comparison with other R packages
Comparison with vanilla PELT
Advanced examples
Custom logistic regression model


Package source: fastcpd_0.14.3.tar.gz
Windows binaries: r-devel:, r-release:, r-oldrel:
macOS binaries: r-release (arm64): fastcpd_0.14.3.tgz, r-oldrel (arm64): fastcpd_0.14.3.tgz, r-release (x86_64): fastcpd_0.14.3.tgz, r-oldrel (x86_64): fastcpd_0.14.3.tgz
Old sources: fastcpd archive


Please use the canonical form to link to this page.