A Study of Applying Sub-Dimensional Approach to k-Means Clustering

• Main Authors: Shu-Mei Cheng, 鄭淑美
• Other Authors: Chien-I Lee
• Format: Others
• Language: zh-TW
• Published: 2004
• Online Access: http://ndltd.ncl.edu.tw/handle/02372725292064626427

碩士 === 臺南師範學院 === 教師在職進修資訊碩士學位班 === 92 === Clustering is one of the most popular methods in data mining. Clustering can rapidly and efficiently create high similar group and decrease the complexity in the same group to continue other technologies of data mining. So clustering is usually used as a preparation before using other data mining functionalities and modularizing. K-Means is one of the most commonly used clustering algorithms. The time complexity of k-Means is O(kmn), which k is the number of clusters, m is the dimensions of data space and n is the total number of objects. The more k, m and n, the more computational time is needed. Therefore, many algorithms are devoted to decrease the time complexity of k-Means. And this thesis is also focused on how to decrease the distance computing between objects and centers, how to reduce the times of comparison, and also decreasing the time complexity of k-Means. For this purpose, decreasing times of comparison between objects and centers by diminishing dimensional computing is applied in this thesis. This way the time complexity of clustering is down from O(kmn) to O(k''m''n), which k''≦k and m''≦m in order to accelerate the clustering.