Random parameter analysis of geometric effects on freeway crash occurrence

摘要: This dissertation presents a comprehensive modeling basis for analyzing the relationship between geometry and crash occurrence on interstates. The analysis is based three empirical papers in ISI indexed journals that investigate total frequencies 1,153 directional interchange non-interchange interstate segments Washington State, USA period 1999 to 2007. conclusions recommendations chapter includes insights from fourth journal paper. It be noted author first all four published articles. paper random parameter negative binomial model of annual crashes by accounting roadway segment heterogeneity marginal effects geometrics. Transportation Research Record, Journal Board US National Academy Sciences shows proof concept geometric database inclusive travel lanes, shoulder widths, information, horizontal vertical alignment, addition average daily traffic, useful as statistical estimation frequencies. Further, article significant models, fixed models are not sufficient. second paper, Safety Science, outlines plausible typology evaluation traffic safety performance. presented integrates network-screening methods with pure outcome methods, multivariate fixedparameter well extensions Highway Manual performance functions. comparative these criteria including, method's integrability network screening techniques, data requirements, predictive importance an overall architecture within which specific type need cast prediction. findings this suggest estimated have promise terms updates functions outlined American Association State Officials document. final Accident Analysis Prevention, extends framework 21 outcomes across typologies including: a) type, b) severity, c) vehicle involvement count, d) collision type. can improved design policy prioritization application via characterization multiple types. In major scientific contributions above, concludes offering integration works, into future research directions. particular, further draw upon article, (in Analytic Methods Research).