An alternative accident prediction model for highway-rail interfaces.

Area of Science:

• Transportation Engineering
• Traffic Safety
• Statistical Modeling

Background:

• Highway-rail interfaces face persistent safety concerns, with high accident frequencies despite improved designs.
• A significant portion of accidents occur at public crossings with active warning systems, indicating a need for better prediction methods.

Purpose of the Study:

• To address limitations in existing highway-rail crossing accident prediction models.
• To develop and present an improved accident prediction model for highway-rail crossings.

Main Methods:

• Utilized negative binomial regression to develop an alternate accident prediction model.
• Evaluated the model's performance against existing methods, considering explanatory variables and prediction accuracy.

Main Results:

• The developed negative binomial regression model offers a simplified, one-step estimation process.
• The new model provides comparable data requirements and clearer interpretation of influential factors.
• Demonstrates promise in improving the accuracy and usability of accident prediction at highway-rail crossings.

Conclusions:

• Existing accident prediction formulas have significant limitations in accuracy and complexity.
• The proposed negative binomial regression model presents a more effective alternative for highway-rail crossing safety analysis.
• This new model enhances the understanding and prediction of factors contributing to highway-rail crossing accidents.