Driving Style Analysis Using Data Mining Techniques

Constantinescu, Zoran; Cristian, Marinoiu; Vlădoiu, Monica · 2010 · OpenAlex-citations

DOI: 10.15837/ijccc.2010.5.2221

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This paper addresses the need for objective methods to classify driver risk-proneness and improve traffic safety, particularly for fleet management. While previous studies relied on subjective expert judgments, questionnaires, or simulators, this research models personal driving styles in real-world urban traffic using data mining techniques. The goal is to categorize drivers based on behavioral indices derived from vehicle telemetry, allowing for the identification of aggressive or unsafe driving patterns. The study utilized the Gipix system, an in-house developed GPS-based tracking device installed in vehicles, to collect raw data including position, time, and speed at one-second intervals. Data was gathered from 23 drivers in Bucharest, Romania, over 2–5 working days, resulting in 200 tracks. Two additional controlled test drives represented extreme styles: one very aggressive (D91) and one slow, economical (D94). From the raw speed data, the authors calculated longitudinal acceleration and mechanical work. Seven specific driving parameters were extracted for analysis: percentage of time exceeding 60 km/h, mean and standard deviation of speed, standard deviation of acceleration, mean and standard deviation of positive acceleration, mean and standard deviation of braking (negative acceleration above a threshold), and total mechanical work. The authors applied Hierarchical Cluster Analysis (HCA) using Ward’s method and Principal Component Analysis (PCA) to the dataset. HCA grouped the drivers into six distinct clusters, successfully isolating the two extreme test drives into separate clusters. PCA revealed that the first principal component correlated strongly with acceleration, braking, and mechanical work, while the second correlated with speed variables. To improve interpretability, the authors performed a varimax rotation, resulting in three rotated components that clearly distinguished acceleration usage, braking style, and speed tendencies. The analysis identified five levels of aggressiveness and differentiated drivers by their speed habits and pedal usage patterns. The findings demonstrate that statistical clustering and PCA can effectively categorize driving styles based on objective telemetry data. The six identified clusters ranged from non-aggressive drivers with smooth braking and moderate speeds to highly aggressive drivers characterized by high speeds, frequent acceleration, and sudden braking. The authors conclude that this approach provides a valuable tool for fleet managers to identify risky drivers and for individuals to assess their own driving safety. They note that future work should incorporate additional factors such as driver demographics, environmental conditions, and real-time video analysis to further refine the models.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• telematics crash prediction
• exposure measurement
• induced exposure

Information type

What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).

• Methodological Resource: dataset resource, tool software
• Theoretical Contribution: computational model