Using Multidimensional Data to Analyze Freeway Real-Time Traffic Crash Precursors Based on XGBoost-SHAP Algorithm

Table 1

Real-time traffic crash precursors obtained from some research results.

Number	Author (year)	Factors affecting the risk of the crash
1	Golob and Recker [11] (2004)	Environmental conditions and traffic flow status
2	Lee et al. [18] (2009)	Average volume, flow variance coefficient, and flow ratio
3	Xu et al. [13] (2010)	Speed standard deviation
4	Christoforou et al. [19] (2011)	Average flow, lane to lane speed variance, average speed, lane-to-lane flow variance, and average density
6	Yu and Abdel-Aty [15] (2014)	Snow season indicators, slope indicators, speed standard deviation, and temperature
7	Wang et al. [20] (2015)	Mainline speed at the beginning of the interweaving section, speed difference at the beginning and end of the interweaving section, and logarithm of the traffic volume
8	Yang et al. [21] (2018)	Upstream average flow, crash section average flow, crash section average speed, and crash section speed standard deviation
9	Yin [22] (2021)	Downstream speed, upstream occupancy, downstream speed coefficient of variation, upstream speed standard deviation, traffic flow status, and time conditions