Latent Clustering Models for Outlier Identification in Telecom Data

Figure 1

Anomaly detection for 5 different models in the sample set defined in Section 4. The three values with the lowest likelihood are circled in orange. Each color represents a different time-stamp class (only 1 class for (a) and (c); 24 classes for (b) and (d)).

(a) Gaussian

(b) Time-Gaussian

(c) GMM

(d) Time-GMM and GPLSA