A hierarchical Bayesian model in Evaluating Apps Downloading Frequency

Several purchase models are developed to understand customer dynamic purchasing behavior. The negative binomial distribution (NBD) contains two characteristics which are Poisson purchasing and gamma heterogeneity is shown as a robust method in product purchase frequency. An approach is to develop individual-level estimation by MCMC Bayesian NBD model. Since the smart phone launch to publish, various types of apps are developed and try to approach users. The app can work successfully only when they meet customers’ requirements. An empirical exam is to estimate effects that influence download frequency for different type of apps. Three app types, game, social communication and photo editor, are examined. The proposed model can capture app download frequency successfully. The game downloaders are more focus on social and hedonic. The social communication downloaders emphasis on informativeness and price. The photo editor downloaders concern on informativeness and social characteristics. Results can provide valuable designing strategies for companies. Key-Words: negative binomial distribution (NBD), MCMC, hierarchical Bayesian model, App