Safety Performance Functions (SPFs) are essential tools to help agencies predict crashes and understand influential factors. This dissertation developed SPFs based on a new Florida Department of Transportation (FDOT) context classification system which contains eight context categories as opposed to three classifications used in the Highway Safety Manual (HSM). Data were first collected for several potential predictor variables based on the Model Inventory of Roadway Elements (MIRE) 2.0, allowing for standard data collection nationwide. Average Annual Daily Traffic (AADT) for minor roads is an important variable in SPF models but it is usually not collected by agencies. Therefore, prior to developing SPF models, a minor AADT model was developed to predict these volumes at intersections without them. Next, SPFs were developed for 19 of the 32 intersection groups (signalized and unsignalized, 3-leg and 4-leg, for each of the eight context classification categories) with sufficient number of intersections and crashes to develop statistically significant models. Multiple modeling methodologies (Negative Binomial (NB), Poisson, Boosted Regression Trees (BRT), Zero-inflated NB (ZINB), and Poisson (ZIP)) were compared using several performance measures to identify the best-performing context-specific SPF for each intersection group. The ZINB models were selected for two intersection groups and NB models were selected for the remaining 17 intersection groups. Each developed SPF had a unique set of significant variables, demonstrating the importance of context-specific SPFs to identify the different influential variables across classification categories. To further show the benefits of the context-specific SPFs, several comparisons were made between individual context-specific SPFs and full SPFs (using data from all four intersection groups within each category) as well as HSM SPFs. The individual context-specific SPFs outperformed full and HSM SPFs, indicating that context-specific SPFs can more accurately predict crash frequencies. These comparisons showed the improved accuracy and additional insights provided by the individual context-specific SPFs.


If this is your thesis or dissertation, and want to learn how to access it or for more information about readership statistics, contact us at STARS@ucf.edu

Graduation Date





Al-Deek, Haitham


Doctor of Philosophy (Ph.D.)


College of Engineering and Computer Science


Civil, Environmental and Construction Engineering

Degree Program

Civil Engineering




CFE0008326; DP0023763





Release Date

December 2025

Length of Campus-only Access

5 years

Access Status

Doctoral Dissertation (Campus-only Access)

Restricted to the UCF community until December 2025; it will then be open access.