Seeded intervals and noise level estimation in change point detection: A discussion of Fryzlewicz (2020)

TL;DR

This paper discusses seeded versus random intervals for change point detection and introduces a new noise level estimator that enhances model selection, especially in low SNR scenarios.

Contribution

It compares interval selection methods and proposes a novel noise level estimator that improves change point detection accuracy in challenging conditions.

Findings

Seeded intervals offer advantages over random intervals in certain contexts.

The new noise estimator improves model selection in low SNR scenarios.

Enhanced change point detection performance demonstrated in challenging cases.

Abstract

In this discussion, we compare the choice of seeded intervals and that of random intervals for change point segmentation from practical, statistical and computational perspectives. Furthermore, we investigate a novel estimator of the noise level, which improves many existing model selection procedures (including the steepest drop to low levels), particularly for challenging frequent change point scenarios with low signal-to-noise ratios.