An unsupervised neural network approach to predictive data mining

Rule mining is one of the popular data mining (DM) methods since rules provide concise statements of potentially important information that is easily understood by end users and are also useful patterns for predictive data mining (PDM). This paper proposes rule mining methods using an unsupervised neural network approach. Two methods are adopted based on the way of unsupervised neural networks are applied in rule mining models. In the first method, the unsupervised neural network is used for clustering, which provides class information to the rule mining process. In the second method, automated rule mining takes the place of trained neurons as it grows in a hierarchical structure of unsupervised neural network.