Classification of Imbalanced Banking Dataset using Dimensionality Reduction

The classification is an important data mining technique. In classification, the classifier automatically learns the properties of classes or categories from the pre-defined training documents. In this paper we have used the imbalanced bank marketing dataset for direct marketing campaign available at UCI machine learning repository. In this paper, we compare the accuracy of different classification algorithms like Naive Bayes, J48, KNN and Bayesnet. CfsSubsetEval method is used for the dimensionality reduction by evaluating the subset of attributes based on the predictive ability and finding out the duplication among the selected features. Before the dimensionality reduction, J48 provide 89%. of accuracy. By using the dimensionality reduction, the accuracy is increased to 91.2% by J48 algorithm.