Abstract
Small and medium sized insurance companies are reducing the need for Big Data analytics mixed AI
Tech., but they are hesitant to adopt Big Data analytics because there are no successful cases. The
domestic financial sector has a large and diverse amount of internal data, with most transactions
occurring in the online environment. Accordingly, Big Data activation level is higher than that of other
industries. This means that the potential value and utilization of data analysis is very high. Accordingly, this
study seeks to develop a model that predicts the effective lapse risk of maintenance contracts. We
conducted an analysis to identify the effective probability of each contract and the effective causes for
defense. In addition, the predictive model was developed using supervised machine learning, and the
effective predictive model was updated by applying unsupervised reinforced learning. By use of Machine
Learning, Reinforcement Learning and Clustering Analysis, we could perform an experiment on lapse
cause mapping to get significant lapse causes