Study on safety risk factors in traffic accidents using data mining techniques = 데이터 마이닝 기법을 이용한 교통사고 위험요인에 관한 연구

Cited 0 time in webofscience Cited 0 time in scopus
  • Hit : 326
  • Download : 0
One of the main topics in traffic safety studies is identifying important causal factors of traffic accidents. Various conventional methods have been widely applied to historical accident data for this purpose. However, recent digitization of historical accident reports and increase of database size have encouraged the application of various data mining techniques to data driven safety researches. This broadened choice of statistical approaches has contributed to the discovery of new knowledge from a large-scale accident database and mitigated weaknesses of existing models stemming from predefined assumptions. This study aims to discover characteristics of traffic accident risk factors by applying various data mining techniques to a large-scale accident database. A historical accident database of the state of California from 2004 to 2010 is used as a data source. This study addresses three subjects in traffic safety: (1) identification of the influential risk factors on injury severity with consideration of interdependency; (2) application of the key risk factors to a logistic regression model for an injury severity analysis, and (3) evaluation of the effectiveness of the handheld cellphone law. The first subtopic applies a few methodologies using two different classification methods, the Naive Bayes Classifier and the Decision Tree Classifier, to identify the relatively important risk factors with respect to injury severity level and analyze dependency among the risk factors. The scope of the analysis is constrained to accidents involving only cars and 25 risk factors in the SWITRS data. The findings are that some important risk factors have strong interdependency and considering the dependency among the top risk factors is vital for an accurate analysis. In the second subtopic, we apply the identified key risk factors to a logistic regression model which is one of the typical methods in an accident severity analysis. The evaluation is performed using the two...
Advisors
Yoon, Yoon-Jinresearcher윤윤진
Description
한국과학기술원 : 건설및환경공학과,
Publisher
한국과학기술원
Issue Date
2014
Identifier
591706/325007  / 020105011
Language
eng
Description

학위논문(박사) - 한국과학기술원 : 건설및환경공학과, 2014.8, [ vii, 100 p. ]

Keywords

Traffic Safety; 교통사고 이력데이터; 위험요인; 데이터 마이닝; 교통안전; Historical Accident Data; Data Mining; Risk Factor

URI
http://hdl.handle.net/10203/197899
Link
http://library.kaist.ac.kr/search/detail/view.do?bibCtrlNo=591706&flag=dissertation
Appears in Collection
CE-Theses_Ph.D.(박사논문)
Files in This Item
There are no files associated with this item.

qr_code

  • mendeley

    citeulike


rss_1.0 rss_2.0 atom_1.0