Edit (2018-12-20):

What i’m talking about here is skimming top performer’s either via outlier’s or those in their respective qualitative classes (such as those with positive residuals in a predictor effect plot)

Initial Post: 2014-10-07

I work on this algorithm for Dwarf Therapist where I try to derive a % fit for a specific Job (Role) per Dwarf; based on a composition of their various variables.

I realized this is what Monte Carlo simulations due to derive probabilities for events (integration).

One could use the higher %’s in further carry forward simulations as a way to direct behavior.

Currently evolution is used to create future successive generations based on performance (survival of the fittest) which is easily extended into AI.