딥러닝기반의 금융회사 고객이탈 예측모형에 관한 연구 : 중소기업 금융에 대한 시사점 도출

At the core of the Fourth Industrial Revolution are new technological innovations in six areas: big data analytics, artificial intelligence, robotics, the Internet of Things, unmanned vehicles, three-dimensional printing, and nano-technology. As a result, the domestic financial environment is changing rapidly. Competitiveness among financial institutions is intensifying by proposing digital electronic banking regulation as a new paradigm. Intensifying competition and the development of information technology have established large-scale databases to create corporate customers, competitors, and new business models. Efforts to effectively analyze and informatize large amounts of data for efficient decision making are increasing. Customers dissatisfaction and abandonment their main bank or company would negatively affect company’s financial performance. so it will be a key point for future management strategies by managing customer satisfaction to attract existing customers and attract new customers. Recent studies in big-data, the artificial intelligence technique were applied to quantify deep learning-based customer's activity type into a measurable variables. Moreover, it is needed for effective and practical survival strategy to overcome these changes. In this study, the optimal model evaluation was performed through the analysis of the results of each methodology of logistic regression, decision tree, and neural networks among the data mining techniques for predicting customer departures. All three models showed good predictions, but the prediction rate of neural network model was the most accurate. In addition, the results were analyzed through an auto-encoder. In the future, it is considered that the methodology applied in this study needs to be applied to policy finance demand forecasting and field of SMEs.