Logistic regression model of pedestrian injury risk

Research output: Chapter in book / Conference proceedingConference article published in proceeding or bookAcademic researchpeer-review

Abstract

This study illustrates the data mining of historical crash records to determine the associations between the risk of mortality or severe injury and all possible contributory factors. Information on the demographic, crash, environment, and traffic characteristics of 73, 746 pedestrian casualties in Hong Kong during the period 1991-2004 was used to establish a predictive model with binary logistic regression. Taking into consideration the impact of temporal interruption on the measures of association, the temporal confounding and interaction effects were progressively incorporated to achieve a more effective predictive model. The goodness-of-fit of the model was assessed using the Hosmer-Lemeshow statistic and a logistic regression diagnostic.
Original languageEnglish
Title of host publicationProceedings of the 11th International Conference of Hong Kong Society for Transportation Studies
Subtitle of host publicationSustainable Transportation
Pages147-155
Number of pages9
Publication statusPublished - 1 Dec 2006
Externally publishedYes
Event11th International Conference of Hong Kong Society for Transportation Studies: Sustainable Transportation - Kowloon, Hong Kong
Duration: 9 Dec 200611 Dec 2006

Conference

Conference11th International Conference of Hong Kong Society for Transportation Studies: Sustainable Transportation
CountryHong Kong
CityKowloon
Period9/12/0611/12/06

ASJC Scopus subject areas

  • Automotive Engineering
  • Civil and Structural Engineering
  • Mechanical Engineering
  • Safety, Risk, Reliability and Quality
  • Transportation

Cite this