Hybrid Prediction Model for the Success of Bank Telemarketing

Abstract

Telemarketing is a technique or system of direct marketing, wherein a businessperson interacts with clients to persuade them to purchase or avail the products and facilities, either by connecting via telephone or through in-person interaction. In the present-day generation, with the humongous acceptance of cellular phones telemarketing has gained popularity as an efficient mode of marketing. In the banking domain, telemarketing is the prime support system to exchange goods and services. Banking products and services promotion to increase the business requires a comprehensive understanding of current market information and the actual client expectations. The present work has investigated traditional data mining and classification approaches which are less accurate. They could not achieve a high customer conversion rate with direct marketing campaigns. The proposed work recommends a machine learning method to foreshow the accomplishments of telemarketing requests for contracting bank term deposits. A Portuguese bank was tagged, considering the impacts of the present economic crisis. The comprehensive set of features linked with bank customer, products and services were inspected. A discussion on four machine learning (ML) models is performed along with the hybrid model, logistic regression ML model (LR), naive Bayes ML model (NB), decision trees ML model (DTs) and support vector machine ML model (SVM). The four ML models were tested and analysed with the proposed hybrid machine learning method (artificial neural network + extreme gradient boosting). The proposed hybrid machine learning method demonstrates the best results.