Anomaly Detection. Time Series AR Testing

This workflow tests the performance of previously trained auto-regressive models for anomaly detection:
- Filter the data to the maintenance window
- Loop over each frequency column
- Apply the previously trained model to the data
- Calculate 1st level alarms based on the prediction errors
- Calculate 2nd level alarms as the moving average of the 1st level alarms
- Visualize the 2nd level alarms in a stacked area chart (all sensors) and a line plot (one sensor)