CAJ | 학술논문

FCM算法作为基于目标函数的模糊聚类算法中最经典的算法之一,在实际应用中得到了深入的研究,但FCM算法需要人为给定分类数C,因此破坏了聚类的无监督性。针对FCM算法的不足,提出了利用密度指标确定初始聚类数目上限Cmax,并且对有效性指标进行了改进,计算对于（1,Cmax]中的每一个c对应的有效性函数值,根据有效性评判,确定最佳聚类数,实现了自动得到最佳分类数的算法。
FCM산법작위기우목표함수적모호취류산법중최경전적산법지일,재실제응용중득도료심입적연구,단FCM산법수요인위급정분류수C,인차파배료취류적무감독성。침대FCM산법적불족,제출료이용밀도지표학정초시취류수목상한Cmax,병차대유효성지표진행료개진,계산대우（1,Cmax]중적매일개c대응적유효성함수치,근거유효성평판,학정최가취류수,실현료자동득도최가분류수적산법。
FCM algorithm as one of the most classic algorithm based on the objective function fuzzy clustering algorithm in practical application,has been deeply researched。But it require the given classification number C,thus damaging the unsupervised cluster.Aiming at the defects of the FCM algorithm,this paper proposed density index for determining the initial cluster the maximum number of Cmax,improved the effective index,calculate every corresponding effectiveness function value of C in the（1,Cmax],and according to the validity judgment determine the optimal cluster number,thus get the best classification number algorithm automatically.