1. Eight numbers were generated, three numbers for each of the inputs recency, frequency, and monetary between 0 and 128. At this point, quiz and first midterm exam membership functions were incorporated into optimization at fixed intervals.
2. Previous input interval values for classical fuzzy and expert fuzzy models were added to population. Each line of the matrix of the initial population, which consisted of a total of 10 x 45 matrixes, represented the intervals of the membership functions for fie inputs.
3. Values in the matrix of the initial population were applied to the fuzzy model and operated. Those values with the lowest error value according to the fitness function were listed. In the present study, the fitness function was error ratio function.