detect.LpS {VARDetect}R Documentation

Single change point detection for low-rank plus sparse model structure

Description

Single change point detection for low-rank plus sparse model using minimizing SSE method.

Usage

detect.LpS(data, lambda, mu, alpha_L, skip = 50)

Arguments

data

A n by p dataset, n is the number of observations, p is the number of variables

lambda

A 2-d vector recording tuning parameters for sparse components left/right

mu

A 2-d vector recording tuning parameters for low rank components left/right

alpha_L

A numeric value, a positive number indicating the constraint space for low rank matrix

skip

The number of observations at the boundaries we should skip

Value

A list object, which includes the followings:

cp

A vector recording all estimated change points

S_hat1

Estimated sparse component for the left-handed side

S_hat2

Estimated sparse component for the right-handed side

L_hat1

Estimated low rank component for the left-handed side

L_hat2

Estimated low rank component for the right-handed side

sse

The sum of squared errors for all passed time points


[Package VARDetect version 0.1.8 Index]