Macro-Level Modeling of Urban Transportation Safety: Case-Study of Mashhad (Iran)

Mohammadi, Mehdi; Shafabakhsh, Gholamali; Naderan, Ali · 2017 · Crossref

DOI: 10.1515/ttj-2017-0025

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 underutilization of proactive safety measures in the planning stage of urban transportation systems. While traffic safety is often managed reactively after crashes occur or through engineering designs, this research aims to integrate crash prediction models into the urban transportation planning process. The primary objective was to evaluate how modal share—the distribution of trips across different vehicle types—impacts crash frequencies at a macro level, using Mashhad, Iran, as a case study. By predicting crashes based on planning data, the authors sought to provide tools for officials to improve safety management proactively. The methodology employed two statistical modeling approaches: Generalized Linear Models (GLM) with a negative binomial distribution and Geographically Weighted Regression (GWR). Data were aggregated at the Traffic Analysis Zone (TAZ) level, encompassing 253 zones in Mashhad. The independent variables consisted of trip volumes by mode (e.g., private cars, buses, taxis, school services) derived from the Mashhad Transportation Master Plan, while the dependent variable was crash frequency from the city’s crash database. The negative binomial model was selected over Poisson regression to account for over-dispersion in crash data. Model development utilized forward selection, validating results through metrics such as the Akaike Information Criterion (AIC), Log Likelihood, and coefficients of determination. Software tools included IBM SPSS Statistics 23 for GLM and ArcMap 10 for GWR. The results demonstrated that both modeling methods were competent in predicting urban crashes. In the negative binomial model, trips by private cars, buses, and bus services were statistically significant at the 95% confidence level. Specifically, private car trips had a positive coefficient, indicating that increased car usage correlates with higher crash frequencies. Conversely, bus trips exhibited a negative coefficient, suggesting that a higher modal share of bus travel reduces crash volumes. The GWR model yielded similar findings, with private cars and school service trips showing positive impacts on crashes, while bus trips remained negatively correlated. The models showed good fit, with the negative binomial model achieving an R² of 0.28 and the GWR model an adjusted R² of 0.40. Visual comparisons of observed versus predicted crashes confirmed the models' accuracy across urban traffic areas. The significance of this research lies in its demonstration that macro-level modeling can effectively support proactive safety planning. The findings imply that shifting modal share from private cars to public buses can significantly improve urban transportation safety. The study concludes that integrating these prediction models into transportation planning allows for more accurate crash forecasting and better-informed policy decisions. It suggests that future research could further refine these models by combining geographically weighted approaches with negative binomial distributions to capture both spatial heterogeneity and count data characteristics.

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
• urban rural setting

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