ALGORITMA GENETIKA DENGAN ROULETTE WHEEL SELECTION DAN ARITHMETIC CROSSOVER UNTUK PENGELOMPOKAN

Abstract

Genetic algorithms are techniques that can be used to clustering data that has global search characters. This application is made using roulette wheel selection techniques and arithmetic crossover techniques. The purpose of this research is to implement a genetic algorithm that produces good results in clustering image data. The result is clustering flower images with different colors has good results, while clustering flower images with similar colors do not have good results. Several experiments were carried out on each scenario to determine the effect of the parameters used on the fitness value obtained, the result was a clustering with parameter color characteristics, the parameter with the largest fitness value are the number of population = 100, iterations = 200, and mutations = 0.02. while clustering with color plus texture characteristic, the parameter with the largest fitness value are the number population=200, iterations=300, and mutations=0.02.