Data-Driven Urban Traffic Accident Analysis and Prediction Using Logit and Machine Learning-Based Pattern Recognition Models

Table 1

Description of variables used in the study.

Variable	Variable levels
Accident severity	1. PDO
	2. Injury/fatal
Accident time	1. 00 : 00 to 06 : 00
	2. 06 : 00 to 12 : 00
	3. 12 : 00 to 18 : 00
	4. 18 : 00 to 24 : 00
Accident day	1. Saturday to Tuesday (major workdays in Iran)
	2. Wednesday to Friday and holidays
Road category	1. Arterial road
	2. Access road
Road surface condition	1. Dry
	2. Humid
	3. Wet
Geometry of accident location	1. Alignment
	2. Roundabout
	3. U-turn/J-turn
	4. Uphill
	5. Downhill
	6. Alley
	7. Intersection
Daylight condition	1. Day
	2. Night
	3. Sunset and sunrise
Accident type	1. Light vehicle-light vehicle
	2. Light vehicle-motorcycle
	3. Light vehicle-heavy vehicle/pickup truck-truck/bus
Type of collision	1. Head-on collision
	2. Rear-end collision
	3. Side-impact collision
Driver gender	1. Male
	2. Female
Driver age	1. Less than 18
	2. 18 to 30
	3. 30 to 45
	4. 45 to 60
	5. 60 and over
Weather condition	1. Clear
	2. Cloudy
	3. Rainy
Reason of accident	1. Lack of attention
	2. Backover movement
	3. Inability to control the vehicle
	4. Exceeding lawful speed
	5. Crossing a forbidden place
	6. Failure to observe longitudinal spacing
	7. Failure to observe lateral spacing
	8. Unsafe lane changes
	9. Hasty-caused accident
	10. Unsafe lane changes
	11. Failure to yield the right-of-way
	12. Improper turns
	13. Wrong way movements
	14. Yaw motion of the vehicle to the left
	15. Technical defect in the vehicle
Vehicle	1. KIA Pride
	2. Paykan
	3. Renault
	4. Peugeot
	5. Taxi
	6. Pickup truck
	7. Minibus
	8. Bus
	9. Truck
	10. Trailer