PPT Slide

• Often non-incremental, bottom-up with similarity measures

• Often incremental,

　　top-down with hill-climbing　search

　Common features of clustering

Data Analysis Machine Learning

• Clusters that optimize

some quality criteria

• Clustering and characterization

are not independent - a

partitioning of data is viewed

as ‘good’ if each cluster has a

‘good’ conceptual description

Example of a cluster:

instances = { c1, c2 }

desc.: #nuclei = 1 & #tails = 2

前のｽﾗｲﾄﾞ

次のｽﾗｲﾄﾞ

最初のｽﾗｲﾄﾞに戻る

ｸﾞﾗﾌｨｯｸｽの表示