DefinePK

Abstract

Data classification is a major task in data mining paradigm. In this paper an artificial neural network approach is proposed for data classification. In this approach data classification is accomplished through a cluster analysis. It is a two-pass process, clusters are created in the first step and classification is achieved from the results of first pass. A self organizing map neural network (SOMNN) is used for clustering in the first pass. In the second pass classification task is completed by using multilayer neural networks (MNN). Basically SOM is an unsupervised neural network and multilayer networks are supervised neural network, hence this approach is a hybrid method. Nine hybrid neural networks (HNN1 to HNN9) are constructed from the combination of above said methods and are experimented. Performance of each hybrid neural network is evaluated by using metrics such as accuracy, precision, recall, and F-measure. Feed back of library users is used as data set for classification.