Diffusion-Based Data-Driven Assortment Optimization

2026-08-11

AI Summary

This work proposes a model-agnostic assortment optimization framework that circumvents the limitations of traditional parametric choice models, which are prone to misspecification and struggle to capture complex customer behavior. The approach introduces, for the first time, a guided discrete diffusion model to this problem by representing assortments as binary vectors and learning a reverse diffusion process to perform stochastic search. A revenue-guided mechanism is embedded within the diffusion dynamics to effectively balance exploration and exploitation. The method reliably generates diverse, high-performing near-optimal solutions in high-dimensional settings, offering both scalability and robustness against model misspecification.

Abstract

Assortment optimization is a fundamental problem in revenue management, typically addressed using parametric choice models such as the multinomial logit (MNL) and its variants. While these models enable tractable formulations, their performance is sensitive to model misspecification and often struggles to capture complex customer behavior. In this paper, we propose a model-agnostic framework for assortment optimization based on guided discrete diffusion. We represent assortments as binary vectors and perform stochastic search via a learned reverse diffusion process, avoiding explicit combinatorial enumeration. To incorporate decision objectives, we introduce a reward-guided mechanism that biases local transitions using estimates of expected revenue. This allows the method to effectively balance exploration and exploitation during generation. Empirically, we show that the proposed approach consistently identifies high-quality assortments and remains robust under model misspecification, often recovering near-optimal solutions in high-dimensional settings. Moreover, the generative nature of diffusion enables the production of diverse high-performing assortments, offering flexibility beyond a single deterministic solution. These results highlight the potential of generative modeling as a scalable and robust paradigm for combinatorial optimization in data-driven decision-making.

Problem

• assortment optimization
• model misspecification
• customer behavior
• revenue management
• combinatorial optimization

Innovation

• diffusion-based optimization
• assortment optimization
• model-agnostic
• reward-guided diffusion
• generative combinatorial optimization