Skip to main content
. 2016 Dec 21;11(12):e0167520. doi: 10.1371/journal.pone.0167520

Fig 2. Cluster analysis.

Fig 2

Agglomerative hierarchical cluster analysis for the whole data set to define the natural division of groups: Pairs of object were defined in the data set (for every subject: duration of intake and TDI score), then a similarity matrix representing every pair of objects of the data set was obtained and a hierarchical cluster tree was built by using the distance information. Clustering then was subsequently conducted by pruning the branches of the hierarchical tree based on the consistency criterion and two subclusters were obtained (cluster 1: marked with dots, cluster 2: marked with crosses).