本文提出一個以資料間的灰關聯分析為基礎的資料聚類方法。由資料間的灰關聯分析結果,我們可以經由一個閥值的設定來判斷資料與資料間的相似關係,再利用這個相似度的測量結果將資料分類。由於閥值的決定關係到資料聚類的結果,在本文中提出一個測量效能的方法來協助選擇適當的閥值。此一具有自動選擇參數的灰聚類演算法由於不需做任何人為的設定,即可以將資料自動聚類,為一非監督式的聚類方法。本文最後會利用幾個例子來說明灰聚類演算法的資料聚類結果。 A data clustering algorithm based on the grey relational analysis ofdata is proposed. From the result of the grey relational analysis, wecan determine the similarity between data by setting a properthreshold value, and the similarity measurement can apply to dataclustering. Since the determination of the threshold value willdirectly affect the result of the clustering, we address a performancemeasurement to select a suitable threshold value. The grey clusteringalgorithm need not a priori setting of parameter, and can clusteringthe data directly. Some examples of data clustering are utilized toillustrate the effective of this algorithm.