WMAPE
Weighted mean absolute percentage error
Error metric AEYS uses to judge how well a trained SEU model predicts energy consumption, shown as a ±% band on predict tables.
Also known asweighted MAPE
WMAPE weights absolute percentage errors by actual consumption so large loads dominate the score. AEYS stores a WMAPE threshold with each saved model and flags rows outside the band as overflow.
Training WMAPE, holdout WMAPE, and live predict overflow all feed SEU performance narratives in the company energy report.
Sources
- AEYS unit management / mf-python models