Anomaly Detection model details

The Model tab for a generated Anomaly Detection model displays information about the peer groups in the model.

Note that the peer group sizes and statistics reported are estimates based on the training data and may differ slightly from actual scoring results even if run on the same data.

The Residual of the unreported reasons is 1 minus the sum of the mean anomaly index values for each of the anomaly columns for the records that are identified as anomalies. This percentage can serve as an indication of how much of the anomaly is explained by the fields that are reported. This might guide you in determining how many anomaly fields to report.