Hierarchical Clustering of Features on Categorical Data of Biomedical Applications

Data mining became increasingly important in bioinformatics and biomedical area during last decade. Various data mining methods, such as association rule mining and clustering, have successfully revealed previous unknown knowledge in bioinformatics and biomedical area. Within the categorical data mining domain, one of the major tasks is to understand the relationship among features. However, none of previous methods addresses the problem of revealing the relationship among the features if a priori knowledge is unknown. In this paper, a previous neglected problem, feature association mining problem is defined. While feature selection method identifies the features associated with the pre-defined target concept, the feature association mining has to be done without any a priori knowledge. To find out the feature association in the categorical data, based on the contingency table, we first define a distance measure to identify the closeness between features. Then, we propose a hierarchic...

• Type: Conference
• Year: 2008
• Where: CAINE
• Authors: Yi Lu, Lily R. Liang
• Keywords: CAINE 2008, Computer Science, Data Mining, Feature Association Mining, Priori Knowledge