Customer retention is critical in the subscription business since it generates a fundamental driving force for the business model. Previous researchers and managers in this organization analyzed customer behavior to prevent the churning event. Our research focuses on Netflix customer pattern analysis with a novel approach, graph theory. The graph methods capture the customer’s movie session with the feature of each movie and predict the link for the next film. Based on IMDB data, we simulated the prediction graph in a movie recommendations environment. The simulation environment recommends a hundred sets of movies to the customer graph model, and the model decides whether it retains the subscription. Our study contributed to the behavioral research using a graph theoretical framework and also the simulation study by utilizing trained graph models as agents.
