Dynamic Assortment Planning Under Nested Logit Models

Dynamic Assortment Planning Under Nested Logit Models
复制标题

嵌套 Logit 模型下的动态分类规划

DOI:
10.1111/poms.13258
复制
发表时间:
2021
影响因子:
5
通讯作者:
Zhou, Yuan
Zhou, Yuan
中科院分区:
管理学3区
文献类型:
--
作者:
Chen, Xi;Shi, Chao;Wang, Yining;Zhou, Yuan

文献摘要

参考文献

被引文献

相似文献

研究了有限长销售季节T内的动态分类规划问题。在每一个时间段,卖方提供一个到达客户的替代产品的分类和客户之间提供的产品根据离散选择模型进行购买。卖方的目标是最大化预期收入,或者等价地,最小化最坏情况下的预期后悔。一个关键的挑战是,产品的效用是未知的卖方,需要学习。虽然动态分类规划问题在收益管理中越来越受到重视,但现有的工作大多是基于多项logit选择模型(MNL)。在本文中,我们研究了一个更一般的选择模型下的动态分类规划问题-嵌套logit模型,该模型模拟分层选择行为,是“GEV(广义极值)家族中使用最广泛的成员”(Train 2009)。通过利用每个嵌套中最优分类的收入排序结构,我们开发了一种新的具有聚合估计方案的置信上限(UCB)策略。我们的策略同时学习客户的选择行为,并根据当前的知识做出动态的决策。它实现了累积的遗憾的顺序,其中M是巢的数量和N是产品的数量在每个巢。我们进一步给出了一个下界结果,表明当T远大于MandN时,上界是接近最优的。当每个nestN的项目数是大的,我们进一步提供了一个离散化的启发式算法更好的性能。数值结果表明,我们所提出的算法的经验性能。
We study a stylized dynamic assortment planning problem during a selling season of finite lengthT. At each time period, the seller offers an arriving customer an assortment of substitutable products and the customer makes the purchase among offered products according to a discrete choice model. The goal of the seller is to maximize the expected revenue, or equivalently, to minimize the worst‐case expected regret. One key challenge is that utilities of products are unknown to the seller and need to be learned. Although the dynamic assortment planning problem has received increasing attention in revenue management, most existing work is based on the multinomial logit choice models (MNL). In this paper, we study the problem of dynamic assortment planning under a more general choice model—the nested logit model, which models hierarchical choice behavior and is “the most widely used member of the GEV (generalized extreme value) family” (Train 2009). By leveraging the revenue‐ordered structure of the optimal assortment within each nest, we develop a novel upper confidence bound (UCB) policy with an aggregated estimation scheme. Our policy simultaneously learns customers’ choice behavior and makes dynamic decisions on assortments based on the current knowledge. It achieves the accumulated regret at the order of, whereMis the number of nests andNis the number of products in each nest. We further provide a lower bound result of, which shows the near optimality of the upper bound whenTis much larger thanMandN. When the number of items per nestNis large, we further provide a discretization heuristic for better performance of our algorithm. Numerical results are presented to demonstrate the empirical performance of our proposed algorithms.
MNL-Bandit:品种选择的动态学习方法
DOI: 10.1287/opre.2018.1832
发表时间: 2017
期刊: ArXiv
影响因子: --
作者:
Shipra Agrawal;Vashist Avadhanula;Vineet Goyal;A. Zeevi
通讯作者: A. Zeevi
MNL-Bandit 的 Thompson 采样
DOI: --
发表时间: 2017
期刊: Annual Conference Computational Learning Theory
影响因子: --
作者:
Shipra Agrawal;Vashist Avadhanula;Vineet Goyal;A. Zeevi
通讯作者: A. Zeevi
DOI: --
发表时间: 2018-10
期刊: J. Mach. Learn. Res.
影响因子: --
作者:
Xi Chen;Yining Wang;Yuanshuo Zhou
通讯作者: Xi Chen;Yining Wang;Yuanshuo Zhou
动态多项 Logit 分类选择模型的近最优策略
DOI: --
发表时间: 2018
期刊: Neural Information Processing Systems
影响因子: --
作者:
Yining Wang;Xi Chen;Yuanshuo Zhou
通讯作者: Yuanshuo Zhou
一种解决潜在类 Logit 分类问题的分支割算法
DOI: 10.1016/j.endm.2010.05.049
发表时间: 2010
期刊: Electron. Notes Discret. Math.
影响因子: --
作者:
I. Méndez;Juan José Miranda Bront;Gustavo J. Vulcano;Paula Zabala
通讯作者: Paula Zabala