91黑料网

Fares Belkhiria,聽a doctoral student at 91黑料网 in the听惭补谤办别迟颈苍驳听area will be presenting his thesis defense entitled:

Three Essays on Precision Retailing Across Scales:
Habits, Recommendation Systems and Synthetic Population

Thursday, June 4, 2026, at 11:00 AM
(The defense will be聽conducted in hybrid mode)

Student Committee Chair: Prof. Laurette Dub茅

Please note:聽Thesis Defences are only open to members of the 91黑料网 community (Students, Professors and Staff) and not the general public. Members of the 91黑料网 community may participate in-person. Due to limited space availability, please contact the PhD office and we will provide you with the room number.

Abstract

Food retail functions as a central choice infrastructure in daily life, shaping health, affordability and well-being under conditions of bounded rationality, limited time, and marketing complexity. Despite this significance, health-oriented retail interventions often produce modest behavioral change. This dissertation examines why such interventions underperform and proposes a framework for precision retailing that integrates behavioral science, artificial intelligence (AI), and synthetic-population analytics to design more effective and equitable strategies. The research adopts the precision retailing paradigm, which treats food-system transformation as requiring cross-disciplinary, data-integrated, and socially responsible innovation. The first essay investigates how multidimensional shopping habits, repetitive, context-dependent patterns of choice across items, categories, and stores shape responsiveness to healthy nudges. Using multi-year loyalty data and machine-learning-based habit measures, the study uncovers strong stickiness of purchases in typically unhealthy categories. It also demonstrates that habitual customers are less likely to respond to healthy eating nudges than habitual ones, revealing both opportunities and risks for public health-oriented retail design. The second essay introduces a Budget-Aware Next Basket Recommender (BANBR) that operationalizes precision retailing as constraint-aware support. The system reframes personalization as a welfare-relevant intervention by reducing cognitive burden while preserving autonomy, affordability, and preference fit. The third essay extends the analysis to the neighborhood level by constructing a synthetic retail ecosystem that combines census-anchored synthetic populations, georeferenced loyalty transactions, neighborhood food-environment healthfulness, and financial capability indicators. This integration produces privacy-preserving mosaic agents that make it possible to examine how healthier food environments are associated with revealed grocery purchasing under different local economic conditions and across shopper segments. The results show that healthy access alone is not enough; healthier retail environments become more behaviorally consequential in specific combinations of neighborhood opportunity, financial capability, and type of shopper. The three essays are presented to move from diagnosis (habit as a mechanism of intervention failure), to prescription (algorithmic tools for constraint-sensitive support), to contextual evaluation (synthetic-ecosystem analysis of multiscale inequality in retail behavior). The thesis advances precision retailing as an empirically grounded framework for improving individual commercial and social outcomes within complex retail systems.