Exploring the Mechanism of Crashes with Autonomous Vehicles Using Machine Learning

Table 3

Variables’ description and distribution.

Variable	No. of crashes	Distribution (%)
TOC (type of collision)
Head-on = 0	11	8.60
Side swipe = 1	20	15.05
Rear end = 2	85	64.52
Broadside = 3	15	11.83
PMAV (the precrash movement of AVs)
Stopped = 0	51	38.71
Moving = 1	80	61.29
PMCV (the precrash movement of conventional vehicles)
Stopped = 0	12	9.16
Moving = 1	119	90.84
VD (vehicle damage)
None = 0	10	7.52
Minor = 1	80	60.07
Mod = 2	35	26.89
Major = 3	5	3.82
DM (precrash driving mode)
Conventional = 0	69	52.69
Autonomous = 1	62	47.31
AL (accident location)
Intersection = 1	62	47.31
Street = 2	46	35.48
Highway = 3	18	13.98
Parking lot = 4	4	3.23
W (weather conditions)
Clear = 1	99	75.57
Cloudy = 2	22	16.79
Raining = 3	6	4.58
Fog/visibility = 4	4	3.05
L (lighting conditions)
Daylight = 1	85	64.52
Dusk-dawn = 2	3	2.15
Dark-street lights = 3	44	33.33
CS (crash severity)
Uninjured = 0	106	80.65
Injured = 1	25	19.35