Prediction of Bank Customer Telemarketing Success Using Data Mining Models

Abstract

Telemarketing has become a popular marketing tool. In a financial crisis, credit may contract in domestic and international markets, forcing banks to turn to customer deposits, including term deposits, as a source of additional funds. In addition, loans are a main source of income for banks. Therefore, a bank manager’s key responsibility is to attract term deposits and generate loans. Here direct marketing enables banks to reach customers who are likely to make deposits and take loans, so identifying such customer groups is a major challenge for banks. Using the proposed model, a bank manager can better conduct direct marketing campaigns and achieve greater customer satisfaction. A machine learning model is built to predict the success rate of bank telemarketing in identifying valued customers. This work describes a data mining approach to extracting valuable information from a dataset from a recent Portuguese bank telemarketing campaign obtained from the UCI KDD Machine Learning Datasets. To validate prediction accuracy, the proposed model is compared with popular classification models including Naïve Bayes, SVM and Decision Tree. All experiments are implemented in WEKA. The experimental results provide useful insights for strengthening banks both financially and functionally.