EVALUATING METHOD FOR GENETIC ALGORITHM

摘要

PROBLEM TO BE SOLVED: To enable evaluation when data on all individuals are not obtained in an evaluation area and to actualize efficient objective evaluation with the small number of evaluations by classifying data for evaluation into fractionated evaluation areas and objectively calculating evaluation values of the individuals. SOLUTION: The data for evaluation of individuals are obtained and classified into fractionated evaluation areas (S1). The evaluation areas are represented by a two-dimensional table and the data for evaluation are classified. Then reference values of models are calculated by the every evaluation area and the values are made to a mean value of all data included in the areas (S2). In this case, the reference value of some evaluation area is the means value of data for evaluation included in areas including its peripheral areas. Then the data for evaluation are so converted (S3) as to equalize the mean values and dispersion of the respective evaluation areas so that the degree of the contribution to an evaluation value of each evaluation area is standardized. Then the evaluation value is obtained from a coefficient for the reference value of each piece of data for evaluation (S4).