Learn R Programming

blockcpd (version 1.0.0)

Change Point Detection for Multiple Aligned Independent Time Series

Description

Implementation of statistical models based on regularized likelihood for offline change point detection on multiple aligned independent time series. It detects changes in parameters for the specified family for the series as group or block. As a reference for the method, see Prates et al. (2021) .

Copy Link

Version

Install

install.packages('blockcpd')

Monthly Downloads

8

Version

1.0.0

License

GPL (>= 2)

Maintainer

Lucas Prates

Last Published

August 12th, 2022

Functions in blockcpd (1.0.0)

plot.blockcpd

Plot for blockcpd object
compute_hausdorff

Hausdorff distance metric
select_frv

Methodology to aid choosing regularization constant
plot.frv

Plot for graphical selection of the constant
rcpd

Sampler for the CPD Block Model
toy_regularization

Implements the regularization functions used in the estimation
compute_rand

Rand Index Function for change point detection
compute_dynseg

Block segmentation using dynamical programming
compute_jaccard

Jaccard's Index metric
compute_symdiff

Symmetric difference metric
check_input

Checks input from caller
compare_model

Compare or evaluate model performance with respect to other model or ground truth
fit_blockcpd

Fits a blockcpd model
confidence_plot

Plot to check reported change-points
compute_hierseg

Block segmentation using hierarchical algorithm