Xie, Z., & Yan, J. (2008). Kernel density estimation of traffic accidents in a network space. Computers, environment and urban systems, 32(5), 396-406.
A proposed network KDE method for traffic accidents further investigates the impacts of using different kernel functions (Gaussian and Quartic), lixel lengths, and search bandwidths. This research focuses on the road network instead of the entire two-dimensional space, making the analysis more meaningful. However, a criterion for statistical significance is needed to identify hot spots.
要得出路段精確座標輸入可能為shapefile等
Network KDE的提出是為了讓KDE專注在路面而不是整的區域的平面
Tola, A. M., Demissie, T. A., Saathoff, F., & Gebissa, A. (2021). Severity, spatial pattern and statistical analysis of road traffic crash hot spots in Ethiopia. Applied Sciences, 11(19), 8828.
This study analyzed road traffic crash (RTC) hot spots using spatial autocorrelation, Getis-Ord Gi*, and Moran’s I to examine spatial patterns. However, it lacks analysis of roadway parameters such as road length and degree of curvature.
Incremental Spatial Autocorrelation (ISA)被用來找出不同距離下的空間自相關，找出最佳距離用來計算GI
GI地區做圖
Alkaabi, K. (2023). Identification of hotspot areas for traffic accidents and analyzing drivers’ behaviors and road accidents. Transportation research interdisciplinary perspectives, 22, 100929.
This study first describes each traffic accident factor. Then, correlation analysis was conducted, followed by logistic regression to filter out non-significant factors. Finally, Incremental Spatial Autocorrelation (ISA) was used to calculate multiscale Global Moran’s I to identify the peak Z-score, and the Getis-Ord method was applied to detect high-risk areas.
The limitations of this research include the small sample size. Additionally, weather conditions were not included, and the policies considered may not be generalizable to other countries.
使用ISA 作為找出最佳距離，過程中會計算global morans
GI熱點做圖
Gedamu, W. T., Plank-Wiedenbeck, U., & Wodajo, B. T. (2024). A spatial autocorrelation analysis of road traffic crash by severity using Moran’s I spatial statistics: A comparative study of Addis Ababa and Berlin cities. Accident Analysis & Prevention, 200, 107535.
The study compares crash data patterns in low and high-income countries. Average Nearest Neighbor Distance (ANND) was used to determine the significance of spatial clustering, while Global and Local Moran’s I were implemented to examine the significance of spatial autocorrelation and to cluster crash severities, respectively. However, the study was limited by the disparity in the timeframe of data between the two countries.