Application of a Hurdle Model with Random Effects to Explore the Relationship between Operational Characteristics and Safety Performance

Abstract

This study used the bus incident data in Victoria, Australia to establish the relationship between operational characteristics and the safety performance of bus operators. A series of count models were investigated to account for methodological challenges, including excess zeros and panel data structure. The empirical results highlighted the different effects operational characteristics had on the risk and prevalence of bus incidents. Operators of smaller size, providing non-route services and operating in regional areas had a lower risk of having any reported incidents compared with larger route operators and operators in areas of higher accessibility. In cases where at least one incident had been reported, incident frequency was higher for operators with higher fleet total travel distance, older fleets and better roadworthy performance (this factor being counterintuitive). Findings from this study provide safety regulators with evidence-driven opportunities to enhance bus safety, including improving incident reporting practices, the establishment of a comprehensive database for heavy vehicle operators, and specific efforts targeted at older fleets.

Enhanced Aggregate Framework to Model Crash Frequency by Accommodating Zero Crashes by Crash Type

“…It is possible that, across spatial units, several crash configurations might have large share of zero crashes. In the traditional multivariate context, when the presence of such zeros needs to be accommodated, researchers employ zero-inflated or hurdle variants such as zero inflated NB model, zero inflated Poisson model, spatiotemporal random effect zero inflated NB model, multivariate random parameters zero-inflated NB regression model, zero inflated hierarchical ordered probit model, and hurdle NB model (10)(11)(12)(28)(29)(30)(31)(32)(33)(34)(35)(36)(37)(38).…”

Application of Association Rule Mining Algorithm based on 5G Technology in Information Management System

“…In the formula: k(z ∪x) represents the number of times that two items appear in the data at the same time; n(x) represents the degree of data correlation; D(z ⇒ x) represents the confidence of z ⇒ x. Based on the above calculation, in order to achieve the balance of financial information, which is the basis of managing financial information, find the most common problems in financial information and determine the relationship policy in financial information [26].…”

“…Most of the current data mining technologies are aimed at deterministic environments, that is, the data to be mined is deterministic data, and there are only two states of deterministic data that exist or do not exist, that is, there is only a difference between 0 and 1.Data mining in uncertain environment is aimed at uncertain data sets, that is, data is no longer 100% certain, but exists with a certain probability [2]. There are many uncertainties in the data, which are mainly divided into the following points: 1) The original data is inaccurate.…”