Refer to the exhibit.  What occurs when a device analyzed by the machine-learning configuration shown has consistently high memory utilization?
A regression model uses average memory utilization as a feature and average CPU utilization as its target. Sustained memory-utilization values become part of the model’s learned input pattern and raise its learned memory-utilization baseline; high memory utilization itself is not the target condition for an incident. FortiSIEM documents regression as predicting a target field from other feature fields.