A new spatial segmentation method to modelling freeway crash frequency

Yu-Chiun Chiou, Lawrence W. Lan, De Cheng Liu, Chiang Fu

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Due to roadway heterogeneity, it is essential to divide a study network into an appropriate number of segments prior to modelling the crash frequency of that network. To this end, this study proposes a genetic algorithm (GA)-based segmentation method to determine the optimal number of spatial segments and then introduces appropriate count models to estimate the crash frequency on the network. The logic of the proposed GA-based segmentation method is to divide a study network into smallest spatial units (say, 1km), and then cluster the neighbouring spatial units with similar crash occurring patterns into a newly-formed segment, keeping the crash occurring patterns among the formed segments as diverse as possible. The empirical results from Taiwan freeway case study has validated the proposed GA-based segmentation method and the results show that the Negative Binomial model with the proposed segmentation method has outperformed over the other two segmentation methods-Fixedlength method and Interchange method.

Original languageEnglish
Title of host publicationProceedings of the 19th International Conference of Hong Kong Society for Transportation Studies, HKSTS 2014 - Transportation and Infrastructure
EditorsZ. Leng, Y. H. Wang
PublisherHong Kong Society for Transportation Studies Limited
Pages117-124
Number of pages8
ISBN (Electronic)9789881581433
StatePublished - 1 Jan 2014
Event19th International Conference of Hong Kong Society for Transportation Studies: Transportation and Infrastructure, HKSTS 2014 - Hong Kong, China
Duration: 13 Dec 201415 Dec 2014

Publication series

NameProceedings of the 19th International Conference of Hong Kong Society for Transportation Studies, HKSTS 2014 - Transportation and Infrastructure

Conference

Conference19th International Conference of Hong Kong Society for Transportation Studies: Transportation and Infrastructure, HKSTS 2014
CountryChina
CityHong Kong
Period13/12/1415/12/14

Keywords

  • Count models
  • Crash frequency
  • Genetic algorithms
  • Optimal spatial segmentation

Fingerprint Dive into the research topics of 'A new spatial segmentation method to modelling freeway crash frequency'. Together they form a unique fingerprint.

Cite this