An improved asymptotic test for the Jaccard similarity index for binary data

Scott H Koeneman; Joseph E Cavanaugh

doi:10.1016/j.spl.2022.109375

Back

An improved asymptotic test for the Jaccard similarity index for binary data

Journal article

Peer reviewed

An improved asymptotic test for the Jaccard similarity index for binary data

Scott H Koeneman and Joseph E Cavanaugh

Statistics & probability letters, Vol.184, 109375

05/2022

DOI: 10.1016/j.spl.2022.109375

View Online

Abstract

For paired binary data, we propose a new asymptotic test of independence for the Jaccard index. As demonstrated, the test offers marked improvements in maintaining nominal Type I error rates, and exhibits higher power when these error rates are comparable.

Association measures

Bernoulli distribution

Binary data

Jaccard index

Multinomial distribution

Details

Title: Subtitle: An improved asymptotic test for the Jaccard similarity index for binary data
Creators: Scott H Koeneman - University of Iowa
Joseph E Cavanaugh - University of Iowa, Biostatistics
Resource Type: Journal article
Publication Details: Statistics & probability letters, Vol.184, 109375
DOI: 10.1016/j.spl.2022.109375
ISSN: 0167-7152
eISSN: 1879-2103
Publisher: Elsevier B.V
Language: English
Date published: 05/2022
Academic Unit: Statistics and Actuarial Science; Biostatistics; Injury Prevention Research Center; Internal Medicine
Record Identifier: 9984227041802771

Metrics

31 Record Views

12 Times Cited - Web of Science