A data mining approach to characterize road accident locations

Kumar, Sachin; Toshniwal, Durga · 2016 · OpenAlex-citations

DOI: 10.1007/s40534-016-0095-5

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 and characterizing high-frequency road accident locations to improve accident prevention measures in India. While traditional statistical models have limitations with large, sparse datasets, the authors propose a data mining approach to uncover hidden correlations between accident attributes and location characteristics. The research is motivated by the high accident rates in India and the need to understand specific geographical and environmental factors that contribute to frequent accidents, rather than just analyzing accident severity. The methodology employs a two-step data mining process on a dataset of 9,640 road accidents recorded between 2009 and 2014 in Dehradun District, Uttarakhand, obtained from the GVK-EMRI emergency service. First, the k-means clustering algorithm was applied to group 87 accident locations into three categories based on frequency counts: high-frequency (HFAL), moderate-frequency (MFAL), and low-frequency (LFAL). The optimal number of clusters (k=3) was determined using the gap statistic. Second, association rule mining using the Apriori algorithm was performed on each cluster to identify significant correlations among attributes such as road type, lighting, vehicle type, and area surroundings. Rules were evaluated using metrics including support, confidence, lift, leverage, and conviction, with a minimum support threshold of 5%. The results reveal distinct characteristics for each frequency category. High-frequency accident locations (8 locations, 28.82% of accidents) are predominantly found on hilly roads with curves and slopes, as well as at highway intersections near markets. These locations are particularly dangerous for two-wheeler accidents, which constitute the majority of incidents in this cluster. Moderate-frequency locations (22 locations, 35.06% of accidents) are identified as dangerous for pedestrian hits, specifically in colonies near local roads and at highway intersections. Low-frequency locations (56 locations, 36.11% of accidents) are scattered throughout the district, and accidents here are generally non-critical. Specific association rules highlighted strong correlations, such as the link between hills, curves, and highways in HFAL, and the association of colonies with pedestrian accidents in MFAL. The significance of this study lies in its demonstration that data mining techniques can effectively extract actionable insights from road accident data that traditional methods might miss. By categorizing locations based on frequency and identifying specific risk factors for each category, the approach provides targeted information for preventive efforts. For instance, infrastructure improvements or traffic management strategies can be specifically tailored to hilly highway curves for two-wheelers or colony areas for pedestrians. This method offers a reliable framework for transportation engineers and policymakers to prioritize safety interventions based on empirical data patterns.

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.

• incidence prevalence

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).

• Empirical Findings: crash risk outcomes