Application of different negative binomial parameterizations to develop safety performance functions for non-federal...

Area of Science:

• Transportation Engineering
• Traffic Safety Research
• Statistical Modeling

Background:

• Safety performance functions (SPFs) are crucial for understanding crash risk, but research predominantly focuses on high-volume roads.
• Non-federal aid system (NFAS) roads, including rural and urban local roads, have unique characteristics (lower speeds, shorter segments) and limited safety research.
• Existing SPFs often do not adequately represent the safety dynamics of NFAS roads due to their distinct features and lower crash exposure.

Purpose of the Study:

• To address the gap in safety research for NFAS roads by developing specific SPFs.
• To investigate and compare the performance of traditional negative binomial (NB) models with zero-favored negative binomial (NB-L) models for NFAS roads.
• To identify optimal model formulations, including variance and dispersion structures, for accurate safety prediction on NFAS roads.

Main Methods:

• Utilized crash, roadway inventory, and traffic volume data from Virginia spanning 2014-2018.
• Applied traditional negative binomial (NB) models and negative binomial-Lindley (NB-L) models, exploring various variance and dispersion structures.
• Evaluated model performance to determine the most suitable approach for predicting crash frequencies on NFAS roads.

Main Results:

• Negative binomial-Lindley (NB-L) models demonstrated superior performance compared to traditional NB models for NFAS roads.
• The selection of an appropriate variance structure and dispersion function further enhanced the predictive accuracy of the NB-L models.
• The study confirms the need for distinct SPFs tailored to the unique characteristics of NFAS roadways.

Conclusions:

• NB-L models provide a more accurate representation of crash frequencies on NFAS roads than traditional NB models.
• Optimizing variance and dispersion structures is essential for maximizing the effectiveness of SPFs on these road types.
• This research provides valuable tools for improving safety management and planning for non-federal aid system roads.