Related Experiment Video
Updated: May 6, 2026
11:21
Methodology for Establishing a Community-Wide Life Laboratory for Capturing Unobtrusive and Continuous Remote Activity and Health Data
Published on: July 27, 2018
8.6K
Pattern Mining of Older Drivers' Driving Behavior Through Telematics-data-driven Unsupervised Learning
Sonia Moshfeghi1, Jinwoo Jang2
1Ph.D. Candidate, Department of Civil, Environmental, and Geoamatics, College of Engineering and Computer Science, Florida Atlantic University, Boca Raton, FL 33431 USA.
IEEE Sensors Journal
|November 5, 2025
Summary
Older drivers aged 65+ exhibit predominantly conservative driving patterns, according to a study using in-vehicle sensors. This research analyzed driving behaviors to identify distinct styles for improved traffic safety insights.
Area of Science:
- Gerontology
- Transportation Safety
- Data Science
Background:
- Aging populations present unique challenges in road safety.
- Older drivers (65+) face increased crash risks and injury severity.
- Understanding age-specific driving behaviors is crucial for safety interventions.
Purpose of the Study:
- To develop a framework for analyzing and clustering older driver behaviors using in-vehicle sensor data.
- To identify distinct driving styles and patterns within the 65+ demographic.
- To leverage advanced machine learning for complex driving data interpretation.
Main Methods:
- Utilized in-vehicle sensor data, including speed, acceleration, braking, RPM, throttle, fuel, engine, and ambient temperature.
- Applied Self-Organizing Maps (SOMs) for data visualization and dimensionality reduction.
- Employed Deep Embedded Clustering (DEC) combined with K-means and agglomerative methods for pattern identification.
Main Results:
- 5x5 grid SOMs effectively visualized multiple driving features simultaneously.
- DEC + K-means and DEC + agglomerative clustering proved effective for determining optimal cluster numbers.
- Clustering analysis revealed two distinct clusters, with the predominant driving style being conservative.
Conclusions:
- The study successfully identified conservative driving patterns as dominant among older drivers (65+).
- The proposed framework and methodologies are applicable to diverse driving features and demographics.
- Findings support broader applications in traffic analysis, driver behavior modeling, and safety research.
