A Fuzzy Clustering Approach to Identify Pedestrians’ Traffic Behavior Patterns

Saeipour, Parisa; Sarbakhsh, Parvin; Salemi, Saman; Bakhtari Aghdam, Fatemeh · 2023 · Crossref

DOI: 10.34172/jrhs.2023.127

archive: archived pipeline: cataloged verified

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

Summary

This study addresses the challenge of identifying homogeneous groups within heterogeneous pedestrian traffic behavior data to improve traffic safety management. Recognizing that pedestrian behavior is complex and often overlaps between categories, the authors aimed to identify hidden behavioral patterns and assess factors associated with higher-risk behaviors. The motivation stems from the need for more targeted interventions and planning, as traditional hard clustering methods may fail to capture the nuanced, overlapping nature of human behavior data. The researchers conducted a secondary methodological study using data from a cross-sectional survey of 600 pedestrians aged 18 and older in Urmia, Iran. Data were collected via the Pedestrian Behavior Questionnaire (PBQ), which measures five domains: adherence to traffic rules, traffic violations, positive behaviors, traffic distraction, and aggressive behaviors. To handle data heterogeneity and overlapping class boundaries, the study employed the Fuzzy C-Means (FCM) clustering algorithm, a soft clustering method that allows data points to belong to multiple clusters with varying membership degrees. The optimal number of clusters was determined using validity indices, including the fuzzy silhouette index, partition entropy, partition coefficient, and modified partition coefficient. Subsequently, multiple logistic regression was used to identify demographic and behavioral predictors of cluster membership. The analysis identified two distinct clusters: a lower-risk cluster comprising 64.33% of participants and a higher-risk cluster comprising 35.66%. The higher-risk cluster exhibited significantly lower mean scores across all PBQ domains, including adherence to rules and positive behaviors, compared to the lower-risk cluster. Logistic regression revealed several significant predictors for membership in the higher-risk cluster. Individuals aged 33 years or younger (OR = 1.92), males (OR = 1.90), those with six or fewer years of education (OR = 1.74), unmarried pedestrians (OR = 3.61), and users of public transportation rather than personal cars (OR = 2.01) were significantly more likely to exhibit higher-risk traffic behaviors. The findings demonstrate that fuzzy clustering is an effective tool for segmenting pedestrian behavior into distinct risk profiles, overcoming the limitations of hard clustering methods in handling complex behavioral data. By identifying specific demographic groups at higher risk, such as younger, unmarried males with lower education levels, the study provides actionable insights for policymakers. These results can inform the development of targeted safety measures, educational campaigns, and infrastructure planning to reduce pedestrian traffic injuries and enhance overall road safety management.

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.

• pedestrian behavior perception
• vru crash typology