Discovery of Driving Behavior Patterns - Smart Information Systems: Computational Intelligence for Real-Life Applications - page 326

distance, we applied agglomerative hierarchical clustering with complete linkage on
the synthetic dataset.
Figure 12.4 illustrates the generated hierarchical cluster trees for both examined
distance measures on the synthetic time series. The first observation to be made is
that RRR perfectly recovers the cluster structure provided by the ground truth, given

(a)

Wave01 YoYo01

Peak01 YoYo03

Wave03 YoYo02

YoYo01 Wave01

YoYo03 Wave03

Peak03 Wave02

Wave02 Peak01

Peak02 Peak02

YoYo02 Peak03

1.9 2 2.1 0.7 0.8 0.9

DTW RRR

(b)

2xWave03 2xYoYo01

2xYoYo03 2xYoYo03

2xPeak03 2xYoYo02

2xWave01 2xWave01

2xYoYo01 2xWave03

2xPeak01 2xWave02

2xWave02 2xPeak01

2xPeak02 2xPeak02

2xPeak03

2xYoYo02

1.6 1.8 2 2.2 2.4 0.2 0.4 0.6 0.8

DTW RRR

Fig. 12.4 Cluster tree (dendrogram) of univariate a and multivariate b synthetic time series (intro-
duced in Fig. 12.3 ) according to the DTWand RRR distance. The x-axis reveals the distance between
the time series being merged and the y-axis illustrates the corresponding name and shape of the
time series