Table 2.
Correct percentage of predicted classification of the Classification and Regression Tree (CART) model.
| Observed | Predicted Percent Correct (%) | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
|
Roadway
Geometry |
A | B | C | D | E | F | G | H | I | X |
| Overall 1 | Straight 2 | Curve 3 | L200 4 | L500 | L800 | R200 | R500 | R800 | Combined 5 | |
| Alert | 68.2 | 27.3 | 86.4 | 86.4 | 90.9 | 77.3 | 90.9 | 68.2 | 54.5 | 86.4 |
| Fatigue | 31.8 | 68.2 | 59.1 | 50.0 | 36.4 | 22.7 | 59.1 | 81.8 | 72.7 | 90.9 |
| BAC 0.02% | 86.4 | 63.6 | 45.5 | 90.9 | 68.2 | 9.1 | 68.2 | 31.8 | 0.0 | 50.0 |
| BAC 0.05% | 36.4 | 0.0 | 81.8 | 0.0 | 50.0 | 77.3 | 36.4 | 72.7 | 68.2 | 54.5 |
| BAC 0.08% | 31.8 | 77.3 | 27.3 | 45.5 | 18.2 | 72.7 | 72.7 | 45.5 | 50.0 | 63.6 |
| Overall Percentage | 50.9 | 47.3 | 60.0 | 54.5 | 52.7 | 51.8 | 65.5 | 60.0 | 49.1 | 69.1 |
1 “Overall” refers to the whole route that does not divide road conditions, including all straight and curve segments. Four independent variables are measured under the whole route: SP_AVG, SP_SD, LP_AVG and LP_SD. 2 “Straight” refers to the straight segments, where four independent variables, i.e., S_SP_AVG, S_SP_SD, S_LP_AVG, and S_LP_SD are measured. 3 “Curve” refers to the curve segments, where four independent variables, i.e., C_SP_AVG, C_SP_SD, C_LP_AVG, and C_LP_SD, are measured. 4 “L200” refers to a curve to the left with a radius of 200 m, where four independent variables, i.e., L200_SP_AVG, L200_SP_SD, L200_LP_AVG, and L200_LP_SD, are measured. 5 “Combined” refers to the variables measured under different segments, including SP_AVG, SP_SD, LP_AVG, and LP_SD under the straight segment, and the same indicators under six curve segments, respectively. A total of 4 + 4 × 6 = 28 independent variables are measured. BAC: blood alcohol content.