This is a past event
The seminar will be held on campus in the University Court Room and via Teams on September 25th from 3pm - 4:15pm.
Join Bowei Chen, a Senior Lecturer (Associate Professor) in Marketing Analytics and Data Science at the Adam Smith Business School of University of Glasgow.
Abstract:
This study presents an explainable machine learning pipeline for bus accident analysis, employing a comprehensive three-stage approach. First, a topic model is utilized to process unstructured accident reports, enabling feature discovery and engineering. Next, a predictive model accurately classifies accident severity. Finally, Shapley Additive Explanations (SHAP) are applied to provide detailed insights into the classification outcomes. The framework is validated using a comprehensive bus accident dataset from a tier-2 city in China, incorporating both structured and unstructured data. The results demonstrate the framework's effectiveness, aligning with previous research while offering new perspectives for accident prevention strategies.
This research advances computational methods in traffic accident analysis and prevention, with a particular emphasis on explainable analytics. By integrating three distinct machine learning algorithms, the framework achieves both high performance and interpretability, making it adaptable to various analytical challenges. This study contributes to the application of big data and machine learning in transportation research and accident analysis, highlighting the critical role of explainable, data-driven approaches in enhancing safety and decision-making processes.
- Speaker
- Bowei Chen
- Hosted by
- University of Aberdeen
- Contact
-
No booking required.