Multi-Source Data Fusion for Intelligent Traffic Accident Risk Prediction

Haitao Huang, Tao Wang, Zandi Shang and Jiandong Cao

GBP Proceedings Series, 2025, vol. 14, 170-178

Abstract: This study addresses the major challenges of traffic accident risk prediction, including pronounced data heterogeneity, intricate spatiotemporal dependencies, and the limited availability of high-risk samples. To overcome these obstacles, it proposes an integrated framework that combines intelligent perception techniques with deep learning models. The research systematically elaborates on the processes of data classification, cleaning, alignment, feature encoding, and unified representation, ensuring the consistency and interpretability of multi-source traffic data. Moreover, attention mechanisms and imbalanced sample optimization strategies are embedded into the network architecture to enhance the model's sensitivity to rare but critical risk scenarios. Model training and comparative experiments based on real-world road operation data demonstrate that the proposed approach substantially outperforms conventional methods in both high-risk classification accuracy and generalization performance. These findings highlight the model's robustness and its promising potential for deployment in intelligent transportation systems and proactive road safety management.

Keywords: traffic accident prediction; multi-source data fusion; deep learning; risk grading; imbalanced data (search for similar items in EconPapers)
Date: 2025
References: Add references at CitEc
Citations:

Downloads: (external link)
https://soapubs.com/index.php/GBPPS/article/view/812/793 (application/pdf)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:axf:gbppsa:v:14:y:2025:i::p:170-178

Access Statistics for this article

More articles in GBP Proceedings Series from Scientific Open Access Publishing
Bibliographic data for series maintained by Yuchi Liu ().