The RSE 2 fitness function
is based on the
Relative Squared Error (RSE).
The RSE 2 component is implemented so that you can choose different
simple models besides the usual target average.
Moreover, the RSE 2 differs from the common RSE in the evaluation of the denominator term,
where instead of using the target output to calculate the difference between
the target and the simple model, the model output is used instead, resulting in a more dynamic,
forever adapting solution space.
The RSE 2 fitness function can be combined
with a cost matrix in order to impose specific constraints on the
In addition, the evolvable
logistic threshold, which is intrinsic to the
logistic regression model, can be adjusted by changing the number of bins.