The purpose of this study is to classify mobile payment (m-payment) users’ behavior and determine the relative importance of influencing factors by using support vector machine and logistic regression. By using survey data of 426 users who had transferred payments frequently in previous one year, classification of users and non-user is estimated by using machine learning classifiers. The findings of the confusion matrix confirm that the accuracy of support vector machine is better than the logistic regression. The research confirms perceived value as the most important predictor of usage behavior through both the models, while other predictors as told by Theory of Acceptance and Use of Technology 2 (UTAUT2) varied slightly in each model. This manuscript provides insights for technology managers who are designing services involving m-payments which ultimately help them with a strategy to better address the users’ forfeiture and switching to other brands.

本研究的目的是利用支持向量机和逻辑回归对移动支付（m-payment）用户行为进行分类，并确定影响因素的相对重要性。通过使用前一年频繁转移支付的426个用户的调查数据，使用机器学习分类器估计用户和非用户的分类。混淆矩阵的结果证实了支持向量机的准确性优于逻辑回归。该研究通过两个模型确认感知价值是使用行为最重要的预测因素，而接受和使用技术理论 2 (UTAUT2) 所告知的其他预测因素在每个模型中略有不同。