Machine Learning Anomaly Scoring - Multiple Tolerance Thresholds

Meter Solution Cloud Service now provides a layered threshold model for Machine Learning (ML)-based validation. A threshold configured directly on a Measuring Component overrides the threshold on its Measuring Component Type; the Measuring Component Type threshold overrides the fallback value in the Master Configuration (D2-MDMMasterConfig).

This allows threshold tuning by commodity and type of meter, improving the balance between identifying valid exceptions and limiting unnecessary work.

Steps to enable and configure

To enable and configure this feature, complete these steps:

1. Set the Anomaly Score Threshold (Fallback) in the MDM Master Configuration.
2. On supported Measuring Component Types, enable Enable ML based Validation (Fallback) and optionally specify Anomaly Score Threshold (Fallback).
3. On supported Measuring Components, enable Enable ML based Validation  and optionally specify Anomaly Score Threshold.

Threshold precedence is: Measuring Component, Measuring Component Type, then MDM Master Configuration.