International Journal of Energy Management: Statistics and Reality – Addressing the Inherent Flaws of Statistical Methods Used in M&V

John M. Avina,
PE, CEM, CEA, CMVP, CxA

 

ABSTRACT

When we use regressions for Option C M&V, we use statistical indicators, such as R2 and CV(RMSE) to determine whether the regression is valid and whether the regression should be used to determine energy and demand savings. Although introduced decades ago, fractional savings uncertainty (FSU), a more complex statistical indicator, is finally becoming known among M&V practitioners. All of these statistical indicators used to qualify Option C regressions are human creations and are not based in reality. This article explains how these indicators are inconsistent, unscientific, arbitrary, and often not well-suited for Option C M&V. The industry needs rules that are simple and that will work for all regressions. In this article, I have presented a set of rules that I believe are understandable to practitioners and will avoid the drawbacks of the current statistical indicator thresholds as recommended by ASHRAE and EVO.

.