Predicting Freeway Traffic Crash Severity Using XGBoost-Bayesian Network Model with Consideration of Features Interaction

Table 2

Variables applied in this research and their values.

Variable categories	Variables	Symbols	Values
Control variable	Severity	SEV	0: Property loss; 1: Injuries; 2: Fatal
	Location	LOC	0: Carriageway; 1: Noncarriageway
	Surface condition	RDC	0: Dry; 1: Wet
	Road alignment	RDA	0: Straight; 1: Not straight
Road factors	Roadway type	LAN	0: Normal section; 1: Special section; 2: Complex node
	Central isolation facilities	CIF	0: W-beam guardrail; 1: Isolation pier; 2: Concrete fence; 3: Metal guardrails; 4: Green belt
	Roadside protection facilities	RSP	0: W-beam guardrail; 1: Isolation pier; 2: Concrete fence; 3: Metal guardrails; 4: Green belt; 5: Border tree; 6: Others
	Geography	GEO	0: Plain; 1: Hills; 2: Mountainous
	Visibility	VIS	0: <50 m; 1: 50∼100 m; 2: 100∼200 m; 3: >200 m
Environmental factor	Light condition	LIG	0: Daylight; 1: Dusk/dawn; 2: Lighting at night; 3: No lighting at night
	Crash time	TIM	0: 0–6 am; 1: 6–12 am; 2: 12–18 (pm); 3: 18–24 (pm)