Safety literature traditionally employs crash frequency models over aggregated data on different spatial scales – micro level (such as segment or intersection) and macro level (such as zone or block) to examine crash occurrence while crash outcome models are employed at the disaggregate level (such as crash or driver record) to examine crash consequences. However, such independent model systems ignore the embedded relationship within data across different resolutions and result in mis-specified models. Recognizing these drawbacks, the current research proposes multiple frameworks for integrating multi-level crash analysis models. Specifically, the proposed frameworks integrate (i) macro and micro level crash frequency models, (ii) aggregate and disaggregate level models to estimate crash frequency by severity, (iii) aggregate and disaggregate level models to jointly estimate crash frequency by crash type and severity, and (iv) macro, micro and disaggregate level models to estimate crash frequency by severity while accounting for hierarchical relationships among the different levels. The frameworks employ econometric building blocks including negative binomial (NB), NB-ordered probit fractional split, multinomial logit and ordered probit models while accommodating for unobserved heterogeneity. The empirical analysis is conducted using data from the City of Orlando, Florida. Several model fit measures, validation exercises and elasticity analysis augment the model analysis. The study results highlighted that all the integrated frameworks showed superior performance relative to the non-integrated (independent) model systems at corresponding analysis resolutions in terms of model fit and predictive performance. The validation exercises also highlighted the superiority of the proposed integrated frameworks. Further, capturing spatial unobserved heterogeneity and random parameter effects improved the performance of the proposed integrated frameworks. The study findings show that the application of the proposed integrated frameworks can allow transportation professionals to adopt policy-based, site-specific, and outcome-specific solutions simultaneously.
| Identifer | oai:union.ndltd.org:ucf.edu/oai:stars.library.ucf.edu:etd2023-1422 |
| Date | 01 January 2024 |
| Creators | Pervaz, Shahrior |
| Publisher | STARS |
| Source Sets | University of Central Florida |
| Language | English |
| Detected Language | English |
| Type | text |
| Format | application/pdf |
| Source | Graduate Thesis and Dissertation 2023-2024 |
Page generated in 0.0136 seconds