Completion Date
Fall 12-15-2021
Document Type
Thesis
Degree Name
Master of Science (MS)
Program or Discipline Name
Analytics
First Advisor
Kayla Jordan
Abstract
Traffic crashes account for most of casualties and injuries worldwide, and there has been growing concerns and studies regarding the contributing factors of traffic crashes. There are many factors causing or related to an occurrence of traffic crash, e.g., land use, traffic flow conditions, driver behavior and weather condition. This paper studied the spatial and temporal distribution of crashes on highway and developed real-time prediction models for crash occurrence. Traffic flow data, weather data, and crash data from multiple data sources were collected and processed to develop the model. Multiple machine learning models, such as SVM model and Decision Tree model, were used as the candidate models. It was found that weather, crash time, and traffic flow shortly prior to the crash occurrence are critical impacting factors for real-time crash prediction. The candidate models have low to moderate sensitivity to predict the crash occurrences due to limited sample size. To use the models in a traffic operations environment, a prediction tool with interactive map could be developed to proactively monitor crash hot spots and prepare staffing and resources for the potential crash occurrences.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 International License.
Recommended Citation
Chen, Y. (2021). Traffic Crash Prediction Using Machine Learning Models. Retrieved from https://digitalcommons.harrisburgu.edu/anms_dandt/1