Louisiana Pedestrian Crash Analysis with Multinomial Logit Model and Bayesian Network

Pedestrians are the most vulnerable users of highway transportation system. While encouraging “Green Transportation”, a concerning fact emerges in the United States: pedestrian deaths are climbing faster than motorist fatalities, reaching nearly 6,000 in 2016 - the highest in more than two decades. In Louisiana, pedestrian fatalities reached 110 in 2015, nearly 15% of total traffic fatalities. In the same year, Louisiana pedestrian fatality rate (pedestrian fatalities per 100,000 population) is 2.18, higher than the U.S. average of 1.67. This paper presents an analysis of Louisiana pedestrian crashes from 2006 to 2015 with the multinomial logit (MNL) and Bayesian networks (BN) models to explore the potential relationship between pedestrian injury severity and a host of factors including pedestrian behavior, demographics, and built environment. The MNL model is utilized to identify the significant factors, and the BN model is structured to reveal probabilistic dependence between pedestrian crash severity and explanatory variables. The results indicate that fatal and severe crashes are closely linked to pedestrians’ alcohol or drugs involvement and older age. The probability of having a fatal or severe injury crash is much higher for pedestrian traveling on roadways away from intersection area (i.e., crossing street or walking along or against roadway). The likelihood of pedestrian crashes resulting in fatality or severe injury increases 49% by walking on unlighted roadways with a speed limit higher than 60 mph at night. The findings of this study show some unique characteristics of pedestrian crashes in Louisiana, which can be useful in selecting the targeted countermeasures.

  • Supplemental Notes:
    • This paper was sponsored by TRB committee ANF10 Standing Committee on Pedestrians.
  • Corporate Authors:

    Transportation Research Board

    ,    
  • Authors:
    • Sun, Ming
    • Sun, Xiaoduan
    • Shan, Donghui
    • Armstrong, Destiny
    • Das, Subasish
  • Conference:
  • Date: 2019

Language

  • English

Media Info

  • Media Type: Digital/other
  • Features: Figures; References; Tables;
  • Pagination: 9p

Subject/Index Terms

Filing Info

  • Accession Number: 01698200
  • Record Type: Publication
  • Report/Paper Numbers: 19-01987
  • Files: TRIS, TRB, ATRI
  • Created Date: Dec 7 2018 9:48AM