Related Experiment Videos
A method to cope with the random errors of observed accident rates in regression analysis
1Road Planning Division, Kanto Regional Bureau, Ministry of Construction, Government of Japan, Tokyo.
Accident; Analysis and Prevention
|August 1, 1989
Summary
This study proposes a method for dividing roads into segments to improve traffic accident analysis. By minimizing random errors in accident rates, regression analysis becomes more reliable for understanding road geometric design impacts.
Area of Science:
- Transportation Engineering
- Traffic Safety Analysis
- Statistical Modeling
Background:
- Road geometric design influences traffic accident rates.
- Accurate analysis requires minimizing random errors in accident data.
- Segmentation of road data is crucial for effective regression analysis.
Purpose of the Study:
- To develop criteria for optimal road segmentation in accident rate analysis.
- To evaluate methods for controlling random errors in segment accident rates.
- To enhance the reliability of regression models linking road design to accidents.
Main Methods:
- Evaluating random error in road segment accident rates based on accident frequency and vehicle-kilometers.
- Proposing and comparing alternative criteria for road segmentation.
- Applying methods to a real-world dataset (Tokyo-Kobe Expressway).
Main Results:
- Random error in accident rates depends on accident counts and traffic volume.
- Segments should have similar and small random errors relative to accident rate variance.
- A recommended criterion for practical road segmentation was identified.
Conclusions:
- Effective road segmentation is key to accurate accident rate analysis.
- Controlling random errors improves the efficiency and reliability of regression models.
- The proposed segmentation criteria offer practical utility for transportation safety studies.