A Recommending System for Mobile Games Using Dynamic Nonparametric Model

We contribute to the literature on recommendation systems by proposing the Dynamic Nonparametric (DNP) model that incorporates a fully Bayesian approach into the traditional dynamic collaborative filtering and hence captures the change of customer preferences in the mobile game industry with higher flexibility. We demonstrate that our DNP model closely predicts game usage patterns with higher accuracy than other widely used benchmarks. We also present various applications that reveal the managerial relevance of our model. Personalized recommendation lists of games generated by the DNP model outperform benchmarks in terms of adoption rate and repeated usage. Our model can also be used to improve the prediction performance of future in-app purchases