TY  - JOUR
T1  - A Hybrid Genetic Algorithm and New Criterion for Determining the Number of Clusters
AU - , Sung-Hae Jun 
JO  - International Journal of Soft Computing
VL  - 1
IS  - 4
SP  - 313
EP  - 318
PY  - 2006
DA  - 2001/08/19
SN  - 1816-9503
DO  - ijscomp.2006.313.318
UR  - https://makhillpublications.co/view-article.php?doi=ijscomp.2006.313.318
KW  - Hybrid genetic algorithm
KW  -new clustering criterion
KW  -decision tree
KW  -the number of clusters
AB  - The study of determining the number of clusters has had an effect on the performance of the clustering result. For example, in the K-means clustering algorithm, its clustering result is affected by the initial K as the number of clusters. But it has been determined by subjectively prior knowledge. Frequently this subjective determination may not be optimal. So, in this study, we proposed an objective method for determining the number of clusters using hybrid genetic algorithm. The initial population of our algorithm was generated by uniform distribution based on decision tree process. We also proposed a new criterion for evaluating the performance of clustering results. In the experiments, we verified our works using data sets from UCI machine learning repository.
ER  -