Local Pattern Detection and Clustering Are There Substantive Differences?

• Published in: Lecture notes in computer science pp. 53 - 70
• Main Author: Höppner, Frank
• Format: Book Chapter Conference Proceeding
• Language: English
• Published: Berlin, Heidelberg Springer Berlin Heidelberg 2005; Springer
• Series: Lecture Notes in Computer Science
• Subjects: Applied sciences; Association Rule Mining; Background Model; Computer science; control theory; systems; Data Density; Data Object; Exact sciences and technology; Information systems. Data bases; Local Pattern; Memory organisation. Data processing; Software; Cluster analysis; Local search; Statistical test; High density; Classification; Very large databases; Knowledge discovery
• ISBN: 9783540265436; 3540265430
• ISSN: 0302-9743; 1611-3349
• DOI: 10.1007/11504245_4

The starting point of this work is the definition of local pattern detection given in [10] as the unsupervised detection of local regions with anomalously high data density, which represent real underlying phenomena. We discuss some aspects of this definition and examine the differences between clustering and pattern detection (if any), before we investigate how to utilize clustering algorithms for pattern detection. A modification of an existing clustering algorithm is proposed to identify local patterns that are flagged as being significant according to a statistical test.