32Minimum statistical requirements for regression baseline emissions models

32  Minimum statistical requirements for regression baseline emissions models

 (1) A regression baseline emissions model for an implementation must, at a minimum, meet the following statistical requirements:

 (a) each relevant variable coefficient must have a t‑statistic that is greater than the value for the 2‑tailed t‑distribution at the 95% confidence level for the number of degrees of freedom in the regression (equivalent to 97.5% confidence in a 1‑tailed distribution);

 (b) the adjusted coefficient of determination (adjusted R2) must be greater than 0.75;

 (c) variation in each relevant variable must affect the dependent variable in a way that can be explained by reference to an effect on either:

 (i) the amount of fuel combusted, or electricity consumed, by equipment within the boundary for the implementation; or

 (ii) the level of industrial process emissions produced by emissions-producing equipment within the boundary for the implementation;

 (d) the residuals of the baseline emissions model must be (confirmed using standard statistical tests and approaches):

 (i) homoscedastic; and

 (ii) normally distributed; and

 (iii) free of auto‑correlation;

 (e) the relative precision of the emissions level predicted by the baseline emissions model for the baseline measurement period for the model (as the case may be) calculated using equation 27 must be within View image in the official source 100%.

  (2) A revised regression baseline emissions model need not be tested for compliance with subsection (1) if:

 (a) it has only been revised to account for a change in the electricity emissions factor (EFElec); and

 (b) the original regression baseline emissions model complied with subsection (1) (or was the subject of a declaration that complies with section 33).