Assortment optimization is a central problem in revenue management and has been studied by our community for more than two decades. However, the lack of a standardized, challenging benchmark has made direct and fair comparisons of algorithms difficult, as researchers often rely on private datasets or randomly generated instances.
To address the gap, we are delighted to share a public benchmark for assortment optimization. This repository (https://github.com/wch444/Assortment-Benchmark) provides hundreds of curated “hard” instances for the mixed MNL and the nested logit models. These instances are systematically generated to challenge simple heuristics (like revenue-ordered assortments) and exact optimal solutions from solvers. More details can be found from the GitHub Repo and Section 5 of our companion paper (https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5671592
Click to follow link." target="_blank" rel="noopener">https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5671592).
This is a community resource and we welcome your collaboration. We invite you to test your algorithms against the benchmark and provide feedback on the format, generation methodology, code base, or documentation. Please feel free to share your thoughts with us at Ningyuan.chen@utoronto.ca. We hope this benchmark will be a step toward more open and reproducible empirical comparisons in our field.
Qing Guo, Saman Lagzi, Chenhao Wang, Ningyuan Chen, Guillermo Gallego, Sumit Kunnumkal, Yao Wang, Li Yu
-------------------------------------------
------------------------------
Ningyuan Chen
Associate Professor
Department of Management, University of Toronto Mississauga
Operations Management and Statistics, Rotman School of Management, University of Toronto
e-mail:
ningyuan.chen@utoronto.caWeb: individual.utoronto.ca/ningyuanchen
------------------------------