| Summary: | 碩士 === 國立成功大學 === 工業與資訊管理學系 === 102 === Clustering analysis is a traditional statistical tool that is used for data classification. While data is often described using crisp numbers, this has some limitations with regard to representing the uncertainties inherent in many situations, and thus so-called fuzzy approaches have been developed to deal with this. An intuitionistic fuzzy set (IFS) is a set of 2-tuple arguments, which are characterized by the properties of membership and non-membership. This paper develop a new clustering analysis method to carry out the data description by IFS. There are two stage in this method. In the calculating state, the similarity method proposed by Liang and Shi (2003) is used to measure the distance between two data points and then combine the most similar data. After data integration, the similarity between the integrated data and the rest of data is assessed, and if this similarity greater than the value of , which is decided by a decision-maker, then a second round of integrated is carried out. This then continues until all of the data is integrated, and then a hierarchical structure of the data cluster can be obtained. In the assessment stage, a modified version of the validity index proposed by Babak (2010) is used to measure the clustering results to find the optimal number of clusters. Comparison of the proposed method with other IFS clustering approaches shows that the method developed in this work achieves similar clustering results, but can return more distinct clusters to the decision-maker, and thus provide more useful information. In summary, this study provides a new IFS hierarchical clustering method that is more flexible than existing approaches and can produce clearer clustering results.
|
|---|
