Anomalies

The CLA can compute an anomaly score. This is essentially the inverse of predictions and represents how unexpected an input record was. Anomaly models are created by specifying the model type as TemporalAnomaly.

Anomaly Score

The actual anomaly score is computed as the fraction of active columns that were not predicted. So if 32 out of the 40 active columns were predicted then the anomaly score would be (40-32)/40=0.2.