Author ORCID Identifier
https://orcid.org/0000-0001-5397-0228
Date Available
12-21-2026
Year of Publication
2024
Document Type
Master's Thesis
Degree Name
Master of Science in Civil Engineering (MSCE)
College
Engineering
Department/School/Program
Civil Engineering
Advisor
Dr. Mei Chen
Abstract
This thesis explores the application of machine learning models to address three distinct transportation problems. In the second chapter, machine learning methods are applied to model annual average daily traffic (AADT). AADT plays a critical role in traffic and safety analysis. The study utilizes XGBoost to model AADT while addressing the imbalanced nature of AADT data using an oversampling method called SMOTE. The oversampling technique is adapted for regression tasks using spectral clustering. The performance of the proposed method demonstrates the effectiveness of the proposed approach. The third chapter focuses on detecting secondary crash narratives using a text-mining approach. Leveraging machine learning to detect secondary crashes is highly cost-effective, as manual review of crash narratives is both time-consuming and expensive. This study applies multiple transformer models to detect secondary crashes, modifying their internal structures to enhance performance. The results validate the effectiveness of this approach. In the fourth chapter, the focus turns to modeling crash counts. Accurate modeling and predicting crash counts are crucial for decision-makers tasked with managing transportation infrastructure and safety analysts. Modeling crash count data faces several challenges, including the excessive number of zeros in the crash count dataset. XGBoost is used for this purpose, and confidence intervals are generated using a novel method called conformal prediction. The results exhibit acceptable error metrics, confirming the reliability of both confidence intervals and predictions.
Digital Object Identifier (DOI)
https://doi.org/10.13023/etd.2024.527
Funding Information
Kentucky Transportation Cabinet Office of Highway Safety, “Incident Management and Secondary Crash”, 2023-2024.
Kentucky Transportation Cabinet Office of Highway Safety, “OHS: Estimating Traffic Volume Using Ubiquitous Probe Vehicle Data”, 2022-2023.
Recommended Citation
Gorji Sefidmazgi, Ali, "Application Of Machine Learning in Transportation Engineering" (2024). Theses and Dissertations--Civil Engineering. 149.
https://uknowledge.uky.edu/ce_etds/149
