Investigating Fractal Characteristics in Crash Trends for Potential Traffic Safety Prediction
Fractal theory is a data mining-based technique that has the ability to find patterns in large and complex datasets. It has been successfully applied in various fields, e.g., in stock market forecast. However, its application in traffic safety analysis has remained largely unexplored. Fractal analysis differs from the traditional time series analysis in that it does not assume data stationarity (i.e., the same mean and variance over time), a property that makes it more appealing for exploring crash trends over time. To benefit from fractal theory, a dataset must exhibit fractal characteristics. This paper investigates the existence of fractal characteristics in crash data. The study made use of the fractal dimension (FD) and Hurst exponent (HE) techniques of fractal theory. FD is a non-integer ratio that provides a statistical index of complexity in the pattern and HE quantifies the tendency of a time series data to strongly regress to the mean. Fractal characteristics exist in the trend whenever FD is greater than 1 and when HE lies between 0 and 1. Crash data from Florida were used to investigate the annual statewide crash frequency trend and the annual crash rate (crash frequency per million entering vehicles) trend at randomly-selected signalized intersections. Both techniques detected the existence of fractal characteristics in the trends. Results of the annual crash rate analysis could be applied in the identification of high-crash intersections. For example, fractal analysis using the HE could be used to predict the crash trend and whether a high-crash intersection would continue to be on future high-crash location lists.
-
Supplemental Notes:
- This paper was sponsored by TRB committee ABJ80 Standing Committee on Statistical Methods.
-
Corporate Authors:
500 Fifth Street, NW
Washington, DC United States 20001 -
Authors:
- Haleem, Kirolos
- Alluri, Priyanka
- Gan, Albert
-
Conference:
- Transportation Research Board 95th Annual Meeting
- Location: Washington DC, United States
- Date: 2016-1-10 to 2016-1-14
- Date: 2016
Language
- English
Media Info
- Media Type: Digital/other
- Features: Figures; References; Tables;
- Pagination: 15p
- Monograph Title: TRB 95th Annual Meeting Compendium of Papers
Subject/Index Terms
- TRT Terms: Crash risk forecasting; Data analysis; Data mining; Fractals; Traffic safety; Trend (Statistics)
- Subject Areas: Data and Information Technology; Highways; I72: Traffic and Transport Planning;
Filing Info
- Accession Number: 01590297
- Record Type: Publication
- Report/Paper Numbers: 16-0494
- Files: TRIS, TRB, ATRI
- Created Date: Feb 16 2016 3:31PM