Data Mining for Potential Customer Segmentation in the Marketing Bank Dataset

Abstract

Direct marketing is an effort made by the Bank to increase sales of its products and services, but the Bank sometimes has to contact a customer or prospective customer more than once to ascertain whether the customer or prospective customer is willing to subscribe to a product or service. To overcome this ineffective process several data mining methods are proposed. This study compares several data mining methods such as Naïve Bayes, K-NN, Random Forest, SVM, J48, AdaBoost J48 which prior to classification the SMOTE pre-processing technique was done in order to eliminate the class imbalance problem in the Bank Marketing dataset instance. The SMOTE + Random Forest method in this study produced the highest accuracy value of 92.61%.