Constant-Factor Algorithms for Revenue Management with Consecutive Stays

2025-06-01

AI Summary

This paper studies network revenue management under continuous resource consumption, motivated by applications such as railway ticketing and hotel booking. It incorporates both accept/reject decisions and choice modeling driven by customer preferences—formulated via attraction models like the Multinomial Logit (MNL) model. We propose the first constant-factor approximation algorithm for this setting, overcoming scalability limitations of classical dynamic programming and linear programming approaches. Our method integrates stochastic online matching, Lagrangian duality analysis, threshold-based admission policies, and resource reservation mechanisms. Theoretically, we establish a tight competitive ratio of $1-1/e approx 0.632$ for the accept/reject model. For the more challenging choice model, we achieve the first constant-factor approximation guarantee of $0.125$, markedly improving upon prior approaches that admit unbounded competitive ratios.

Technology Category

• Planning, Routing, and Scheduling: Scheduling under Uncertainty
• Machine Learning: Online Learning & Bandits
• Reasoning under Uncertainty: Stochastic Optimization

Application Category

• Economics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems
• Graph Algorithms and Modeling for the Web: Algorithms and analysis for incomplete, noisy, or partially observed Web-related graphs
• User Modeling, Personalization and Recommendation: User modeling for targeted and personalized online advertising

Abstract

We study network revenue management problems motivated by applications such as railway ticket sales and hotel room bookings. Request types that require a resource for consecutive stays sequentially arrive with known arrival probabilities. We investigate two scenarios: the reject-or-accept scenario, where the request can be fulfilled by any available resource, and the choice-based scenario, which generalizes the former by incorporating customer preferences through basic attraction models. We develop constant-factor approximation algorithms: $1-1/e$ for the reject-or-accept scenario and $0.125$ for the choice-based scenario.

Problem

• Develop algorithms for revenue management with consecutive stays
• Address reject-or-accept and choice-based customer scenarios
• Provide constant-factor approximations for both scenarios

Innovation

• Constant-factor approximation for revenue management
• Handles consecutive stays with known probabilities
• Incorporates customer preferences via attraction models