The Effect Analysis of Crossover and Selection Methods on the Performance of GenClust++ Algorithm

In the K-Means algorithm, the determination of the center point coordinates (centroid) directly affects the quality of the clustering process. Determination of the coordinates of the center point (centroid) is generally done by generating random numbers and each instance will then be placed based on proximity to random numbers generated. Determining a good centroid will prevent the occurrence of local optima problems in K-Means. The GenClust++ algorithm is a pretty good algorithm in determining centroid. However, what needs to be considered is the influence of the selection and crossover process on the performance of clustering. This study will discuss the combination of selection and crossover processes that are good in the process of determining the centroid. Performance measurement method will be based on the measurement of Mean Square Error and distance calculation using Euclidean Distance. The results show that the Roulette Wheel Selection and Whole Arithmetic Crossover selection processes will give the best performance for the GenClust++ Algorithm.