Telemarketing outcome prediction using an ensemble-based machine learning technique

Business organisations often use telemarketing, which is a form of direct marketing strategy to reach a wide range of customers within a short time. However, such marketing strategies need to target an appropriate subset of customers to offer them products/services instead of contacting everyone as people often get annoyed and disengaged when they receive pre-emptive communication. Machine learning techniques can aid in this scenario to select customers who are likely to positively respond to a telemarketing campaign. Business organisations can use their CRM-based customer information and embed machine learning techniques in the data analysis process to develop an automated decisionmaking system, which can recommend the set of customers to be communicated. A few works in the literature have used machine learning techniques to predict the outcome of telemarketing, however, the majority of them used a single classifier algorithm or used only a balanced dataset. To address this issue, this article proposes an ensemble-based machine learning technique to predict the outcome of telemarking, which works well even with an imbalanced dataset and achieves 90.29% accuracy.