Integrated Associative Classification and Neural Network Model Enhanced by Using a Statistical Approach

摘要

Association rules is a novel data mining technique that has been mainly used for data description, exploration and prediction in knowledge d iscovery and decision support systems. The association rule mining algorithm is modified to handle the user-defined input constraints. Associative classification is provided with a large number of rules, from which a set of quality ru les are chosen to develop an efficient classifier. Many attribute selection measures are used to reduce the number of generated rules. In this paper the pruning of rule sets is facilitated by chi squared analysis thereby only positively correlated rules are used in the classifier. Also the Neural Network Associative Classification system is used in order to improve the accuracy of the classifier. The trained network is then used to classify the future data. The performance of the Neural Network Associative Classification system is analyzed with the datasets from UCI machine learning repository