A Hybrid Method for Traffic Incident Duration Prediction Using BOA-Optimized Random Forest Combined with Neighborhood Components Analysis

Table 2

The relevant variables of traffic incident duration selected in this study.

No.	Variables	Descriptions
1	Incident type	1: Accidents, 2: Breakdowns, 3: Debris/Pedestrian
2	Incident time	1: A.M., 2: P.M.
3	Direction	1: Northbound, 2: Southbound
4	Incident location	1: Lane, 2: Right Shoulder, 3: Central Divide
5	Number of lanes closed	1: No; 2: One, 3: More than one
6	Distance from the exit	1: < 0.5 miles, 2: 0.5~1 miles, 3: >1 miles
7	Relative position to exit	1: At the exit, 2: Front of the exit, 3: Behind the exit
8	Detection type	1: Call report, 2: Operator detected
9	Need Police	1: Yes, 2: No
10	Need Ambulance	1: Yes, 2: No
11	Need Truck Wrecker	1: Yes, 2: No
12	Need firefighters	1: Yes, 2: No
13	Weather information	1: Sunny, 2: Cloudy, 3: Light Rain, 4: Heavy Rain
14	Automobile Count	1: None; 2: One, 3: Two; 4: More than two
15	Motorcycle Count	1: None; 2: One, 3: More than one
16	Heavy Truck Count	1: None; 2: One, 3: More than one
17	Light Truck Count	1: None; 2: One, 3: More than one
18	Tractor Trailer Count	1: None; 2: One, 3: More than one
19	Duration	1: <10min, 2: 10~30min, 3: 30~60min, 4: 60~90min; 5: >90min