【Topic】Analysis of Path Data in Marketing with Applications to Grocery Shopping

【Speaker】Professor Eric T. Bradlow

【Time】2007-6-5 10:00-11:00

【Venue】Room120Shunde Building

【Language】English

【Organizer】Department of Marketing, China Business Research Center

【Target Audience】Faculty, Ph.D, Master students

【Background Information】

Summary

Path data, i.e., records of consumers’ movements over time, contain valuable information for marketing researchers because they describe how consumers interact with their environment and make dynamic choices. This research focuses on the analysis of path data, with particular emphasis on the study of a novel data set – integrated grocery shopping path andpurchasing (scanner) data obtained from grocery carts affixed with Radio Frequency Identification tags. The research stream is comprised of three interrelated papers, each of which studies paths using a different approach. In the first paper, we identify the different dimensions and components of a path model and provide a unifying framework that allowsus to apply tools developed in other disciplines (e.g., models of birds, pedestrians, and traffic) to path data in marketing. In the second paper, we develop a stochastic model for store shopping trips to capture the relationship between consumers’ shopping paths through the store and their purchasing behavior. We calibrate our model using data from Sorensen Associates, an in-store research company, and discuss our empirical findings and managerial implications. In the third paper, we view grocery shopping trips through the lens of the “Traveling Salesman Problem,” a classic optimization framework in operations research. Wecalculate the optimal (shortest travel distance) path for each grocery shopper and compare each observed path with its optimal counterpart. We then decompose the systematic deviations from optimality into “order deviation” and “travel deviation”, and associate these deviations to issues regarding store design, product category purchasing decisions, and consumer segmentation.