An evaluation of performance measures for hotspot identification. Issue 4 (1st October 2016)

• Record Type: Journal Article
• Authors: Manepalli, U. R. R.; Bham, G. H.
• Is Part Of: Journal of transportation safety & security. Volume 8:Issue 4(2016)
• Page Start: 327
• Page End: 345
• Publication Date: 2016-10-01
• Subjects: hotspots -- hotspot identification methods -- crash severity -- crash frequency -- crash rate -- EPDO -- empirical Bayes -- network screening; Transportation -- United States -- Safety measures -- Periodicals; Transportation -- Security measures -- United States -- Periodicals; Transportation -- Safety measures; Transportation -- Security measures; United States; Periodicals; 353.9805
• DOI: 10.1080/19439962.2015.1048015
• Languages: English
• ISSNs: 1943-9970

Abstract

ABSTRACT: An important step in highway safety analysis is the identification of high-crash locations or "hotspots." Various performance measures are currently used for hotspot identification (HSID). Current measures are mostly based on crash frequency. This article proposes a performance measure based on crash severity, namely, the crash factor measure (CFM). In this article, the performance measures are classified broadly by crash frequency and crash severity. The crash severity measures, the CFM, the equivalent property damage only (EPDO), and the Empirical Bayes supplemented with crash severity (EBCS), and the crash frequency measures, the crash rate (CR) and the Empirical Bayes (EB) are evaluated using eight tests based on empirical and simulated data. To ensure detailed assessment, the evaluation was carried out for three entire highway routes (an Interstate, a US, and a State highway), that is, beyond the top 1% to 10% of hotspots commonly examined. The results from the evaluation demonstrate that the EB and CFM measures performed consistently and are recommended for use. Based on inconsistent results, the CR measure is not recommended for HSID. Based on the overall results, it is also recommended that, for any proposed performance measure, the evaluation should be carried beyond 1% to 10% of the hotspots.