Authors: Theera Prompreing; Kattareeya Prompreing; Genesis Sembiring Depari; Jen-peng Huang
Addresses: Rajamangala University of Technology Lanna, 41/1 M.7 T. Maingam Mueang Ngam District, Tak Province, Thailand ' Business and Management of Southern Taiwan University of Science and Technology, No. 1 Rd., Nantai Street, Yongkang District, Tainan City, Taiwan ' School of Business, Universitas Pelita Harapan Medan, Jl. Imam Bonjol No. 6, Medan City, Indonesia ' Department of Information Management, Southern Taiwan University of Science and Technology, No. 1 Rd., Nantai Street, Yongkang District, Tainan City, Taiwan
Abstract: In order to keep competing in a competitive market, it is important to have an accurate prediction of which customers are most likely to buy products or services. Data mining method is one of the useful techniques that experts use to deal with this problem. However, there are a bunch of algorithms that can be employed. The question of which algorithm should be used is still a hot issue today. This research aims to find the best machine learning algorithm in predicting telemarketing success, especially for targeting potential customers. We examined eight machine learning algorithms such, deep neural network (deep learning), naive Bayes, generalised linear model, logistic regression, decision tree, random forest, support vector machine, and gradient boosted tree along with adaptive parameters to each of the algorithms. The results show that, gradient boosted trees outperform the other seven algorithms which achieve 91.3% accuracy.
Keywords: bank telemarketing; machine learning algorithm; random forest; gradient boosted tree.
International Journal of Business Information Systems, 2022 Vol.40 No.3, pp.399 - 414
Received: 20 May 2019
Accepted: 25 Dec 2019
Published online: 17 Aug 2022 *