Abstract:Today, North American governments are more willing to consider compact neighborhoods with increased use of sustainable transportation modes. Bicycling, one of the most effective modes for short trips with distances less than 5km is being encouraged. However, as vulnerable road users (VRUs), cyclists are more likely to be injured when involved in collisions. In order to create a safe road environment for them, evaluating cyclists' road safety at a macro level in a proactive way is necessary. In this paper, different generalized linear regression methods for collision prediction model (CPM) development are reviewed and previous studies on micro-level and macro-level bicycle-related CPMs are summarized. On the basis of insights gained in the exploration stage, this paper also reports on efforts to develop negative binomial models for bicycle-auto collisions at a community-based, macro-level. Data came from the Central Okanagan Regional District (CORD), of British Columbia, Canada. The model results revealed two types of statistical associations between collisions and each explanatory variable: (1) An increase in bicycle-auto collisions is associated with an increase in total lane kilometers (TLKM), bicycle lane kilometers (BLKM), bus stops (BS), traffic signals (SIG), intersection density (INTD), and arterial-local intersection percentage (IALP). (2) A decrease in bicycle collisions was found to be associated with an increase in the number of drive commuters (DRIVE), and in the percentage of drive commuters (DRP). These results support our hypothesis that in North America, with its current low levels of bicycle use (<4%), we can initially expect to see an increase in bicycle collisions as cycle mode share increases. However, as bicycle mode share increases beyond some unknown 'critical' level, our hypothesis also predicts a net safety improvement. To test this hypothesis and to further explore the statistical relationships between bicycle mode split and overall road safety, future research needs to pursue further development and application of community-based, macro-level CPMs.

Analysis of the severity of vehicle-bicycle crashes with data mining techniques

Analyse vehicle–pedestrian crash severity at intersection with data mining techniques

Assessing crash severity of urban roads with data mining techniques using big data from in-vehicle dashcam

Leveraging Connected Vehicle Data for Near-Crash Detection and Analysis in Urban Environments

Analysis of crash injury severity on two trans-European transport network corridors in Spain using discrete-choice models and random forests

Data mining approach to model bus crash severity in Australia

Integrating machine learning into path analysis for quantifying behavioral pathways in bicycle-motor vehicle crashes

Analyzing Factors and Interaction Terms Affecting Urban Fatal Crash Types Based on a Hybrid Framework of Econometric Model and Machine Learning Approaches

Exploring injury severity of bicycle-motor vehicle crashes: A two-stage approach integrating latent class analysis and random parameter logit model

Analyzing Accident Injury Severity Via an Extreme Gradient Boosting (xgboost) Model

Identification of Significant Factors in Fatal-Injury Highway Crashes Using Genetic Algorithm and Neural Network

Identifying factors related to pedestrian and cyclist crashes in ACT, Australia with an extended crash dataset

Analysis of Factors Affecting the Severity of Automated Vehicle Crashes Using XGBoost Model Combining POI Data

A hybrid approach for modeling bicycle crash frequencies: Integrating random forest based SHAP model with random parameter negative binomial regression model

Predicting and factor analysis of rider injury severity in two-wheeled motorcycle and vehicle crash accidents based on an interpretable machine learning framework

An empirical tool to evaluate the safety of cyclists: Community based, macro-level collision prediction models using negative binomial regression

Analysis of factors influencing the degree of accidental injury of bicycle riders considering data heterogeneity and imbalance

Older Pedestrian Traffic Crashes Severity Analysis Based on an Emerging Machine Learning XGBoost

A novel feature extraction model for traffic injury severity and its application to Fatality Analysis Reporting System data analysis

Enhancing Prediction and Analysis of UK Road Traffic Accident Severity Using AI: Integration of Machine Learning, Econometric Techniques, and Time Series Forecasting in Public Health Research