factorcpt (version 0.1.2)

common.null.stat: Threshold selection for the common components

Description

Implements the stationary bootstrap procedure, which generates thresholds that are used for multiple change-point detection in the common components. It is internally called for by common.seg and inherits all its arguments.

Usage

common.null.stat(mat, p.seq, B, q, lam, f, scales, mby, tby, do.parallel)

Arguments

mat

a matrix containing the nodes of a binary tree grown for change-point analysis from the estimated common component

p.seq

reciprocals of the average block size for the factors

B

the number of bootstrap samples

q

the number of factors

lam

a matrix containing the estimated loadings

f

estimated factors

scales

see scales in factor.seg.alg

mby

see dmby in func_dc_by

tby

see dtby in func_dc_by

do.parallel

see do.parallel in factor.seg.alg

Value

B values approximating the test statistics under the stationarity over the segments of consideration in the binary tree (defined by the nodes contained in mat).

References

M. Barigozzi, H. Cho and P. Fryzlewicz (2016) Simultaneous multiple change-point and factor analysis for high-dimensional time series, Preprint.