Latent Clustering Models for Outlier Identification in Telecom Data

Figure 2

Identified clusters for 2 models in the sample set defined in Section 4 between 0:00 and 7:00. In (a), each class of one hour contains 5 clusters, and clusters are not related across hours. In (b), the whole set contains 5 clusters.

(a) Time-GMM

(b) GPLSA