Learn R Programming

LinearDetect (version 0.1.5)

Change Point Detection in High-Dimensional Linear Regression Models

Description

A unified framework for simultaneous structural break detection and parameter estimation in high-dimensional linear models. The proposed method can handle a wide range of models, including change-in-mean model, multiple linear regression model, Vector auto-regressive model and Gaussian graphical model.

Copy Link

Version

Install

install.packages('LinearDetect')

Monthly Downloads

66

Version

0.1.5

License

GPL-2

Maintainer

Yue Bai

Last Published

March 22nd, 2021

Functions in LinearDetect (0.1.5)

BIC threshold for final parameter estimation

lm.first.step.blocks

Threshold block fused lasso step for linear regression model.

Generate the gaussian graphical model data with break points

ggm.first.step.blocks

Threshold block fused lasso step for gaussian graphical model.

ggm.second.step.search

Exhaustive search step for gaussian graphical model.

lm.second.step.search

Exhaustive search step for linear regression model.

lambda_warm_up_lm

lambda warm up for linear regression model

BIC and HBIC function

BIC.threshold.ggm

BIC threshold for final parameter estimation (GGM)

remove.extra.pts

helper function for detection check

Generate the linear regression model data with break points

soft threshold function

Prediction function for VAR 2

Prediction function for VAR (block)

constant.sim.break

Generate the constant model data with break points

Plot the cross-validation score

var.first.step.blocks

Threshold block fused lasso step for linear regression model.

Generating non-stationary ARMA data.

var.second.step.search

Exhaustive search step

Threshold block fused lasso (TBFL) algorithm for change point detection

prediction function

Prediction function (block)