type TStepwiseQualityCriteria = function(const Stat: TRegStats; const b, t: TVec; const ParamMask: TVecInt; const aOwner: TObject): double;
Callback function for custom quality criteria of stepwise regression optimization algorithm.
The Stat contains the details of the regression results, b contains the coefficients and t the corresponding t-Values for each item in b. The ParamMask.Length equals the initial total variable count. b and t can and will have length less than this. Which variables are included is determined from ParamMask, which has 1 at corresponding included variable index and 0 otherewise. aOwner is a custom optional object and will be nil if not specified as a parameter to the optimization.