CVRMSE
A building energy consumption prediction validation error metric recommended by ASHRAE, computed as RMSE divided by the observed mean to yield a dimensionless relative error, facilitating cross-comparison across buildings and energy consumption magnitudes. ASHRAE Guideline 14 specifies calibration thresholds of approximately 15% for monthly models and approximately 30% for hourly models.
Timeline (last 90 days)
Compared with existing methods, DySENSE improves prediction performance by 8.7% (accuracy) and reduces prediction error by 9%–18% (CVRMSE, ASHRAE standard).
CVRMSE (coefficient of variation of the root mean square error) is the error metric recommended by ASHRAE for validating building energy consumption predictions, calculated as RMSE divided by the mean of observed values, and is a dimensionless relative error.
All Facts (2)
Compared with existing methods, DySENSE improves prediction performance by 8.7% (accuracy) and reduces prediction error by 9%–18% (CVRMSE, ASHRAE standard).
50%UnverifiedCVRMSE (coefficient of variation of the root mean square error) is the error metric recommended by ASHRAE for validating building energy consumption predictions, calculated as RMSE divided by the mean of observed values, and is a dimensionless relative error.
50%