Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in [1].

Learning Strategies in Decentralized Matching Markets under Uncertain Preferences

Authors: Xiaowu Dai, Michael I. Jordan

JMLR 2021 | Venue PDF | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We conduct simulation studies (Section 5) to assess the properties of CDM and compare the payoffs with alternative methods. We demonstrate aspects of the theoretical results through experiments (Section 6) using real data from college admissions and simulated graduate school admissions.
Researcher Affiliation Academia Xiaowu Dai EMAIL Department of Economics University of California, Berkeley, CA 94720-1776, USA. Michael I. Jordan EMAIL Division of Computer Science and Department of Statistics University of California, Berkeley, CA 94720-1776, USA.
Pseudocode Yes Algorithm 1 Calibrated decentralized matching (CDM) under the mean calibration. [...] Algorithm 2 Calibrated decentralized matching (CDM) under the maximin calibration.
Open Source Code No No explicit statement or link regarding the availability of source code for the methodology described in this paper was found.
Open Datasets Yes In this section we present an analysis of admissions data from the New York Times The Choice blog (https://thechoice.blogs.nytimes.com/category/admissions-data).
Dataset Splits Yes The training data are simulated by having agents pull random numbers of arms according to their latent utilities. The training data consists of 20 rounds of random proposals under each of the arms preference structures. In total, the sample size is T = 200. The testing data draws a random state from {s1, . . . , s10} and generates the corresponding arms preferences. [...] The training data consists of 20 draws of random proposals under each of the students preference structures. In total, the sample size T = 200. The testing data draws a random state from {s1, . . . , s10}, which gives the corresponding student preferences.
Hardware Specification No The paper does not provide specific details about the hardware used for running its experiments.
Software Dependencies No The paper does not provide specific ancillary software details with version numbers.
Experiment Setup Yes Each agent has the same quota q = 5 and the same penalty γ chosen from {2, 2.5, 3}. [...] Each has the same quota q = 5 and penalty γ = 2.5.