make.tree: Growing a binary tree for change-point analysis

In factorcpt: Simultaneous Change-Point and Factor Analysis

Description

Grows a binary tree of a given height for input panel data, with possible changes in the means of component series, using Double CUSUM statistics, where each node contains information about estimated change-points

Usage

make.tree(y, dw, rule = NULL)

Arguments

y	a matrix containing each component series as its rows
dw	see dw in factor.seg.alg
rule	see rule in factor.seg.alg

Value

tree	a list representing a binary tree grown as an outcome from the Double CUSUM Binary Segmentation algorithm; each node is a 6-by-1 matrix containing the index of the node, the start and end of the segment and an estimated change-point candidate, and the test statistic
mat	a matrix stacking the nodes of tree as its columns

Author(s)

Haeran Cho

References

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

factorcpt documentation built on May 2, 2019, 8:15 a.m.