A data mining approach for bank telemarketing using the rminer package and r tool

Due to the global financial crisis, credit on international markets became more restricted for banks, turning attention to internal clients and their deposits to gather funds. This driver led to a demand for knowledge about client’s behavior towards deposits and especially their response to telemarketing campaigns. This work describes a data mining approach to extract valuable knowledge from recent Portuguese bank telemarketing campaign data. Such approach was guided by the CRISP-DM methodology and the data analysis was conducted using the rminer package and R tool. Three classification models were tested (i.e., Decision Trees, Naïve Bayes and Support Vector Machines) and compared using two relevant criteria: ROC and Lift curve analysis. Overall, the Support Vector Machine obtained the best results and a sensitive analysis was applied to extract useful knowledge from this model, such as the best months for contacts and the influence of the last campaign result and having or not a mortgage credit on a successful deposit subscription.