Driver and Path Detection through Time-Series Classification

Table 3

, precision, recall, accuracy, and training time obtained by using the MLP classifier for different windows sizes.

Average window size	Driver	Driver recall	Driver precision	Overall precision	Overall recall	Accuracy	Training time (sec)
1 s; 14592 training segments	DF1	0,66	0,68	0,68	0,68	0,93	1048
	DM1	0,71	0,66
	DM2	0,69	0,7
	DM3	0,65	0,69
10 s; 1459 training segments	DF1	0,76	0,87	0,90	0,90	0,90	564
	DM1	0,86	0,81
	DM2	0,98	0,93
	DM3	0,98	0,98
30 s; 486 training segments	DF1	0,69	0,91	0,82	0,83	0,82	177
	DM1	0,7	0,76
	DM2	0,92	0,85
	DM3	0,97	0,78
60 s; 243 training segments	DF1	0,25	0,94	0,49	0,53	0,44	78
	DM1	0,21	0,26
	DM2	0,51	0,54
	DM3	0,98	0,36
Start&Stop; 7562 training segments	DF1	0,96	0,82	0,96	0,94	0,94	762
	DM1	0,86	0,94
	DM2	1	0,99
	DM3	1	1