Table 7.

Weights of each performance metric in the All group, generated from the expert comparisons using the linear regression model constrained for positive value coefficients (LM Perc)

Metrics	Segment 2	Segment 3	Segment 4	Segment 5	Segment 6
1. CN of EG and VO	0.020	0.004	0.007	0.008	0.010
2. CC of EG and VO	0.027	0.051	2.2e$-$06	0.024	0.080
3. CN of EG	1.8e$-$11	0.014	5.8e$-$08	0.004	0.002
4. CC of EG	7.3e$-$10	0.016	0.025	3.4e$-$10	3.3e$-$11
5. No. of blinks/s	0.011	4.1e$-$10	3.4e$-$10	4.1e$-$10	0.014
6. Screen time	2.2e$-$11	4.8e$-$10	0.017	0.005	4.9e$-$12
7. No. of collisions/s	1.6e$-$10	2.0e$-$11	6.3e$-$09	0.042	0.018
8. Path with speed Perf.	0.674	0.467	0.848	0.768	0.423
9. Path performance	6.0e$-$12	3.5e$-$11	2.0e$-$10	0.011	1.6e$-$12
10. CN of HP and VO	0.016	0.004	0.014	0.015	0.015
11. CC of HP and VO	0.039	0.097	0.002	1.4e$-$10	0.058
12. CN of HP	0.013	4.3e$-$10	6.2e$-$08	0.008	0.018
13. CC of HP	0.008	2.5e$-$12	0.005	0.033	2.4e$-$11
14. CC of user inputs	0.027	0.044	3.5e$-$08	7.2e$-$07	0.048
15. CN of user inputs	0.021	0.047	0.008	0.012	0.015
16. Average braking	2.6e$-$11	0.038	1.1e$-$07	0.009	0.050
17. Average throttle	0.124	0.151	0.075	0.047	0.244
18. Average steering	0.014	0.026	2.4e$-$08	0.014	3.2e$-$12
19. Eye fixations	0.004	0.042	8.8e$-$07	1.2e$-$10	0.006

Table shows the amount of significance of a metric according to the segment’s path. Each column adds up to one. These values were used to derive our results in the expert model

CN cluster number, CC cluster centres, VO virtual orientation, EG eye gaze, HP head pose