Discussion of 'Detecting possibly frequent change-points: Wild Binary Segmentation 2 and steepest-drop model selection'

TL;DR

This paper discusses theoretical guarantees of the WBS2.SDLL method for change-point detection and explores an alternative MOSUM-based approach for candidate generation, aiming to improve detection accuracy.

Contribution

It introduces an alternative MOSUM-based candidate generation method for SDLL and analyzes its theoretical properties compared to existing approaches.

Findings

WBS2.SDLL has proven theoretical guarantees.

MOSUM-based candidate generation offers a viable alternative.

Comparison indicates potential improvements in change-point detection accuracy.

Abstract

We discuss the theoretical guarantee provided by the WBS2.SDLL proposed in Fryzlewicz (2020) and explore an alternative, MOSUM-based candidate generation method for the SDLL.