Per-event significance indicator to visualise significant events

TL;DR

This paper introduces a new visualization method for significant events in searches for new phenomena, based on probability density functions, applicable to both binned and unbinned data, offering an alternative to traditional signal-to-background ratios.

Contribution

It proposes an alternative significance indicator using probability density functions, improving visualization and applicability in unbinned data analyses.

Findings

Reproduces $ ext{log}(s/b)$ for small signal-to-background ratios

Applicable to unbinned data where signal-to-background is ambiguous

Provides a clearer visualization of significant events

Abstract

In this note, an alternative for presenting the distribution of `significant' events in searches for new phenomena is described. The alternative is based on probability density functions used in the evaluation of the `significance' of an observation, rather than the typical ratio of signal to background. The method is also applicable to searches that use unbinned data, for which the concept of signal to background can be ambiguous. In the case of simple searches using binned data, this method reproduces the familiar quantity $\log(s/b)$ , when the signal to background ratio is small.