Accident Prediction Model Based on Macroscopic Traffic Characteristics

Please enable javascript in your browser.

Mohamadian Amiri, Amir | 2011

611 Viewed

Type of Document: M.Sc. Thesis
Language: Farsi
Document No: 41662 (09)
University: Sharif University of Technology
Department: Civil Engineering
Advisor(s): Nassiri, Habibollah
Abstract:
Lately, there has been a great attention to accident prediction methods in traffic management. The importance of predicting an accident is because of its enormous effect on reducing casualties, injuries, property damages and delays. The main objectivein this study was to identify the relationship between accidents frequency rates and geometric design variables, climatic parameters and traffic characteristics using probabilistic models. The other objective was to evaluate the crash severity and crash type. Therefore, the 1997 traffic accident statistics of city of Mashhad in Iran was used for the for the modeling of different highway classification. Probabilistic models are used for evaluating accident frequency rates and logit function is used to predict crash types. Results show thatnegative binomial model is the appropriate model for assessing accident frequency rates. Poisson model did not provide a satisfactory fit mainly due to the equality of its mean and variance. Logit function showed good results in predicting different crash types
Keywords:
Stochastic Model ; Logit Model ; Accident Prediction ; Macroscopic Properties

No TOC