Designing Privacy Aware Dynamic Pricing Optimization Algorithm for Personalized Retail Offers in US Digital Commerce

Abstract

This paper proposes a privacy-aware dynamic pricing optimization algorithm tailored for personalized retail offers in the context of U.S. digital commerce. With the rapid rise of AI-driven surveillance pricing practices and real-time price adjustments, consumers are facing heightened concerns regarding their privacy and pricing fairness. The algorithm, named PrivacyGuard Pricing, integrates advanced machine learning models, including a novel application of Federated Learning combined with Reinforcement Learning (RL), to optimize pricing dynamically while ensuring consumer privacy through data anonymization and decentralized learning. The primary contribution of this work is a privacy-preserving approach to realtime pricing, overcoming the trade-off between personalized offers and privacy by minimizing consumer data exposure. Performance comparisons with traditional pricing models (e.g., ElasticNet regression-based pricing, A/B testing approaches, and neural network pricing models) demonstrate superior accuracy, fairness, and consumer trust. A series of real-world market simulations and experimental results substantiate the algorithm's efficacy, highlighting its potential to disrupt existing pricing systems while adhering to evolving privacy regulations.