Using a machine learning model specifically trained on your meter's historical usage and local weather patterns, we generate daily usage estimates—from a meter’s early data all the way through one year into the future. This includes backfilling gaps in historical usage where data may be missing and providing a comparative or expected value when reviewing Outliers.
How It works:
- Daily Transformation: Monthly meter data is converted into daily usage values.
- Anomaly Detection: We remove usage readings that appear inaccurate or anomalous.
- Weather Integration: We use historical and forecasted weather specific to the meter’s building.
- ML-Driven Estimates: Machine learning models generate expected daily usage values.
- Quality Checks: We ensure all estimates are within realistic bounds—no negative values and no unexpected spikes