Long-term deposits prediction: A comparative framework of classification model for predict the success of bank telemarketing

Abstract

The long-term deposits product is often offered to prospective customers by the bank through telemarketing strategy. How to maximize customer value through telemarketing strategy is a major focus in this research. Therefore, required a model that can classify potential customers with the potential to increase corporate earnings. The Decision Tree (DT), Naïve Bayes (NB), Random Forest (RF), K-Nearest Neighbour (K-NN), Support Vector Machine (SVM), Neural Network (NN), and Logistic Regression (LR) model have been proposed to compare. The comparison of these algorithms evaluated using the Protestal Bank dataset of the UCI Machine Learning repository, for performance algorithms using Area Under Curve (AUC) and Accuracy. From the experiment, the results show that SVM yield promising results with the Accuracy and AUC 97.07% and 0.925 respectively. It can be concluded that SVM is the best choice for classifying prospective customers who have the potential to be interested in time deposit products that are offered by telephone or cellular as distinguished from other classification algorithms.