Research Article

Driver and Path Detection through Time-Series Classification

Table 4

, precision, recall, accuracy, and training time obtained by using the RF and DTW classifiers for different windows sizes.

Average window sizeDriver RandomForest (RF)DTW
Driver recallDriver precisionOverall precisionOverall recallAccuracyTraining time (sec)Driver recallDriver precisionOverall precisionOverall recallAccuracyTraining time (sec)

1 s
14592 training segments
DF10,840,820,830,850,842280,810,720,710,770,72151
DM10,730,870,810,77
DM20,810,890,750,81
DM30,880,890,760,83

10 s
1459 training segments
DF10,760,870,870,880,89480,730,850,830,870,8148
DM10,840,800,810,78
DM20,960,900,910,87
DM30,930,970,890,96

30 s
486 training segments
DF10,670,890,790,800,80210,650,870,750,770,7621
DM10,670,740,630,70
DM20,880,830,840,81
DM30,920,760,890,71

60 s
243 training segments
DF10,240,930,470,510,52150,210,890,440,490,5215
DM10,190,260,170,25
DM20,480,520,430,47
DM30,970,350,940,34

Start&Stop
7562 training segments
DF10,780,780,890,890,851920,740,730,850,870,82192
DM10,830,910,790,91
DM20,960,960,910,94
DM30,990,920,950,88