Estimation of service value parameters for a queue with unobserved balking
Estimation of service value parameters for a queue with unobserved balking
Authors: Daniel Podorojnyi, Liron Ravner Date: 2026-04-17 Paper ID: openalex:2409.04090
Summary
This paper addresses the estimation of latent customer service values in a queueing system where only queue lengths are observed, and balking behavior is hidden. The authors propose a Maximum Likelihood Estimator for the parametric distribution of service values and prove its consistency and asymptotic normality. This framework is further applied to develop a dynamic pricing scheme that optimizes revenue by iteratively refining price updates based on estimated parameters. The proposed methodology is validated through numerical simulation experiments.
Key Contributions
- Constructed a Maximum Likelihood Estimator (MLE) to infer the parametric distribution of customer service values in a queueing system with unobserved balking.
- Established consistency and asymptotic normality for the proposed MLE under verifiable conditions based on observed queue length dynamics.
- Developed a dynamic pricing algorithm that iteratively updates admission prices to maximize revenue by leveraging the estimated service value parameters.
Open Questions & Future Work
Archivist Review
I have reviewed the submission and decided to approve the two open questions provided, as they identify significant, unresolved research bottlenecks regarding the structural limitations of parametric queueing models. I have rejected all concept candidates as the proposed methodology (MLE for queueing parameters) is a standard application of statistical inference rather than a novel, reusable ML architecture or algorithmic pattern deserving of a permanent vault entry.
Approved Open Questions
- Non-parametric estimation for unobserved balking: This addresses the restrictive nature of parametric assumptions in strategic queuing models, potentially expanding applicability to real-world scenarios where underlying utility distributions are unknown.
- Joint estimation of valuation and patience: Relaxing the homogeneity of delay sensitivity is critical for accurately modeling real-world queuing systems where customers have diverse time constraints.
Links
Metadata & Links
- url
- https://arxiv.org/abs/2409.04090
- paper_id
- 2409.04090
- paper_source
- openalex
- domain
- time-series
- tags
-
- architectures
-
- datasets
-
- concept_slugs
-
- dataset_slugs
-
- skill
- GeneralMLSkill
- created_at
- 2026-04-20T05:54:28Z