The amount of data businesses are collecting about their customers is staggering. Firms can now easily track and record past purchases, product usage patterns, and customers’ responses to marketing campaigns and promotion programs. If fully analyzed, such rich transaction data offers companies the opportunity to understand what drives customers’ purchase decisions, how to improve their shopping experience, and how to develop and retain loyal customers. My dissertation addresses these issues by applying consumer analytics, including association rule mining, survival analysis, econometrics, and optimization, on large-scale transactional data to help companies better understand, predict, and subsequently influence the consumption behavior of their customers.

My dissertation comprises three essays. The first essay utilizes multi-level association rule mining to predict project-oriented purchases. In the second essay, I propose an Expo-Decay proportional hazard model and use customers’ adoptions and usage of previous product generations to predict their upgrade behaviors for the current product generation. In the third essay, a time-based dynamic synchronization policy is applied for the maintenance of consolidated data repository under an infinite planning horizon. In these essays, I apply and extend a variety of business analytics tools including data mining (association rule mining and collaborative filtering), survival analysis, dynamic programming, simulation, and econometric models. These essays contribute to the consumer analytics literature and can help firms maintain high-quality data assets and make informed decisions on cross-generation product development, product promotion and recommendation, and customer retention.