An Adjustable Bisecting Clustering Method

• Main Authors: Chi-Che Pan(Peter), 潘麒全
• Other Authors: W.P. Lee
• Format: Others
• Language: zh-TW
• Published: 2003
• Online Access: http://ndltd.ncl.edu.tw/handle/z6n3kq

碩士 === 中原大學 === 資訊管理研究所 === 91 === In knowledge discovery and data mining, clustering is one of the important issues. Its main feature is 『A cluster is a collection of data objects that are similar to one another within the same cluster and dissimilar to the objects in other clusters.』Clustering is useful in a number of different areas, including Social Science, genetics, bioscience, business and education. Two of the most popular methods among many clustering methods are Hierarchical clustering and Partitional clustering. Hierarchical clustering is often portrayed as the better quality clustering approach, but limited because of its quadratic time complexity. Partitional clustering’s advantage is lower time complexity. In past, researcher offer the clustering algorithm called Bisecting K-mean clustering that both have the advantage of Hierarchical clustering that have a better cluster qulity and Partitional clustering that is lower time complexity. But this method’s main drawback is that when some data have divided into wrong cluster in the beginning, they can’t be adjusted in following step due to it’s bisecting step. Therefore this research focus on this “mistake”, and offer a method called Adjustable Bisecting K-mean clustering, this method use “aggregation” to improve the result of Clustering quality and let the “mistakes” can be adjusted in following step. At last, we propose an explanation to verify that the feasibility of this method.