Dynamic Assortment Optimization with Changing Contextual Information

Authors: Xi Chen, Yining Wang, Yuan Zhou

JMLR 2020 | Venue PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental The effectiveness of the proposed policy is further demonstrated by our numerical studies. ... In this section, we present numerical results of our proposed MLE-UCB algorithm. We use the greedy swapping heuristics (Algorithm 4) as the subroutine to solve the combinatorial optimization problem in Eq. (19). We will also study the quality of the solution of the greedy swapping heuristics.
Researcher Affiliation Academia Xi Chen EMAIL Leonard N. Stern School of Business New York University New York, NY 10012, USA; Yining Wang EMAIL Warrington College of Business University of Florida Gainesville, FL 32611, USA; Yuan Zhou EMAIL Department of Industrial and Enterprise Systems Engineering University of Illinois at Urbana-Champaign Urbana-Champaign, IL 61801, USA
Pseudocode Yes Algorithm 1: The MLE-UCB policy for dynamic assortment optimization with changing features ... Algorithm 2: Approximate combinatorial optimization, the univariate (d = 1) case, and with the designated maximum utility item. ... Algorithm 3: Approximate combinatorial optimization, the univariate (d = 1) case. ... Algorithm 4: A greedy heuristic for combinatorial assortment optimization
Open Source Code No License: CC-BY 4.0, see https://creativecommons.org/licenses/by/4.0/. Attribution requirements are provided at http://jmlr.org/papers/v21/19-1054.html. This text refers to the license and attribution requirements for the paper itself, not for any accompanying source code.
Open Datasets No The unknown model parameter θ0 Rd is generated as a uniformly random unit d-dimensional vector. The revenue parameters {rtj} for j [N] are independently and identically generated from the uniform distribution [0.5, 0.8]. For the feature vectors {vtj}, each of them is independently generated as a uniform random vector v such that v = 2 and v θ0 < 0.6.
Dataset Splits No The paper describes generating synthetic data for numerical studies but does not explicitly mention standard training/test/validation splits. It mentions generating "1000 such test instances" for evaluating the greedy swapping algorithm, but this refers to instances of the optimization problem, not a split of a larger dataset for model training and evaluation.
Hardware Specification No The paper describes the experimental setup for numerical studies but does not specify any hardware details such as GPU models, CPU models, or cloud computing resources used.
Software Dependencies No The paper mentions implementing algorithms and numerical studies but does not provide specific software dependencies or their version numbers (e.g., programming languages, libraries, frameworks).
Experiment Setup Yes Experiment setup. The unknown model parameter θ0 Rd is generated as a uniformly random unit d-dimensional vector. The revenue parameters {rtj} for j [N] are independently and identically generated from the uniform distribution [0.5, 0.8]. For the feature vectors {vtj}, each of them is independently generated as a uniform random vector v such that v = 2 and v θ0 < 0.6. Here we set an upper bound of 0.6 for the inner product so that the utility parameters utj = exp{v tjθ0} are upper bounded by exp( 0.6) 0.55. We set such an upper bound because if the utility parameters are uniformly large, the optimal assortment is likely to pick very few items, leading to degenerated problem instances. In the implementation of our MLE-UCB algorithm, we simply set ω = p d log(T).