A multivariate spatial approach to model crash counts by injury severity

Kun Xie, Kaan Ozbay, Hong Yang

Research output: Contribution to journalArticle

Abstract

Conventional safety models rely on the assumption of independence of crash data, which is frequently violated. This study develops a novel multivariate conditional autoregressive (MVCAR) model to account for the spatial autocorrelation of neighboring sites and the inherent correlation across different crash types. Manhattan, which is the most densely populated urban area of New York City, is used as the study area. Census tracts are used as the basic geographic units to capture crash, transportation, land use, and demo-economic data. The specification of the proposed multivariate model allows for jointly modeling counts of various crash types that are classified according to injury severity. Results of Moran's I tests show the ability of the MVCAR model to capture the multivariate spatial autocorrelation among different crash types. The MVCAR model is found to outperform the others by presenting the lowest deviance information criterion (DIC) value. It is also found that the unobserved heterogeneity was mostly attributed to spatial factors instead of non-spatial ones and there is a strong shared geographical pattern of risk among different crash types.

LanguageEnglish (US)
Pages189-198
Number of pages10
JournalAccident Analysis and Prevention
Volume122
DOIs
StatePublished - Jan 1 2019

Fingerprint

Spatial Analysis
Wounds and Injuries
Censuses
Autocorrelation
Economics
Safety
deviant behavior
Land use
urban area
census
land use
Specifications
ability
economics
Values

Keywords

  • Bayesian method
  • Multivariate conditional autoregressive model
  • Safety analysis
  • Spatial statistics

ASJC Scopus subject areas

  • Human Factors and Ergonomics
  • Safety, Risk, Reliability and Quality
  • Public Health, Environmental and Occupational Health

Cite this

A multivariate spatial approach to model crash counts by injury severity. / Xie, Kun; Ozbay, Kaan; Yang, Hong.

In: Accident Analysis and Prevention, Vol. 122, 01.01.2019, p. 189-198.

Research output: Contribution to journalArticle

Xie, Kun ; Ozbay, Kaan ; Yang, Hong. / A multivariate spatial approach to model crash counts by injury severity. In: Accident Analysis and Prevention. 2019 ; Vol. 122. pp. 189-198.
@article{43d5f8a74a1d4fe1af1b530687cef45f,
title = "A multivariate spatial approach to model crash counts by injury severity",
abstract = "Conventional safety models rely on the assumption of independence of crash data, which is frequently violated. This study develops a novel multivariate conditional autoregressive (MVCAR) model to account for the spatial autocorrelation of neighboring sites and the inherent correlation across different crash types. Manhattan, which is the most densely populated urban area of New York City, is used as the study area. Census tracts are used as the basic geographic units to capture crash, transportation, land use, and demo-economic data. The specification of the proposed multivariate model allows for jointly modeling counts of various crash types that are classified according to injury severity. Results of Moran's I tests show the ability of the MVCAR model to capture the multivariate spatial autocorrelation among different crash types. The MVCAR model is found to outperform the others by presenting the lowest deviance information criterion (DIC) value. It is also found that the unobserved heterogeneity was mostly attributed to spatial factors instead of non-spatial ones and there is a strong shared geographical pattern of risk among different crash types.",
keywords = "Bayesian method, Multivariate conditional autoregressive model, Safety analysis, Spatial statistics",
author = "Kun Xie and Kaan Ozbay and Hong Yang",
year = "2019",
month = "1",
day = "1",
doi = "10.1016/j.aap.2018.10.009",
language = "English (US)",
volume = "122",
pages = "189--198",
journal = "Accident Analysis and Prevention",
issn = "0001-4575",
publisher = "Elsevier Limited",

}

TY - JOUR

T1 - A multivariate spatial approach to model crash counts by injury severity

AU - Xie, Kun

AU - Ozbay, Kaan

AU - Yang, Hong

PY - 2019/1/1

Y1 - 2019/1/1

N2 - Conventional safety models rely on the assumption of independence of crash data, which is frequently violated. This study develops a novel multivariate conditional autoregressive (MVCAR) model to account for the spatial autocorrelation of neighboring sites and the inherent correlation across different crash types. Manhattan, which is the most densely populated urban area of New York City, is used as the study area. Census tracts are used as the basic geographic units to capture crash, transportation, land use, and demo-economic data. The specification of the proposed multivariate model allows for jointly modeling counts of various crash types that are classified according to injury severity. Results of Moran's I tests show the ability of the MVCAR model to capture the multivariate spatial autocorrelation among different crash types. The MVCAR model is found to outperform the others by presenting the lowest deviance information criterion (DIC) value. It is also found that the unobserved heterogeneity was mostly attributed to spatial factors instead of non-spatial ones and there is a strong shared geographical pattern of risk among different crash types.

AB - Conventional safety models rely on the assumption of independence of crash data, which is frequently violated. This study develops a novel multivariate conditional autoregressive (MVCAR) model to account for the spatial autocorrelation of neighboring sites and the inherent correlation across different crash types. Manhattan, which is the most densely populated urban area of New York City, is used as the study area. Census tracts are used as the basic geographic units to capture crash, transportation, land use, and demo-economic data. The specification of the proposed multivariate model allows for jointly modeling counts of various crash types that are classified according to injury severity. Results of Moran's I tests show the ability of the MVCAR model to capture the multivariate spatial autocorrelation among different crash types. The MVCAR model is found to outperform the others by presenting the lowest deviance information criterion (DIC) value. It is also found that the unobserved heterogeneity was mostly attributed to spatial factors instead of non-spatial ones and there is a strong shared geographical pattern of risk among different crash types.

KW - Bayesian method

KW - Multivariate conditional autoregressive model

KW - Safety analysis

KW - Spatial statistics

UR - http://www.scopus.com/inward/record.url?scp=85055681923&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85055681923&partnerID=8YFLogxK

U2 - 10.1016/j.aap.2018.10.009

DO - 10.1016/j.aap.2018.10.009

M3 - Article

VL - 122

SP - 189

EP - 198

JO - Accident Analysis and Prevention

T2 - Accident Analysis and Prevention

JF - Accident Analysis and Prevention

SN - 0001-4575

ER -