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].

A Study of Proxies for Shapley Allocations of Transport Costs

Authors: Haris Aziz, Casey Cahan, Charles Gretton, Philip Kilby, Nicholas Mattei, Toby Walsh

JAIR 2016 | Venue PDF | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We perform an experimental evaluation using synthetic Euclidean games as well as games derived from real-world tours calculated for scenarios involving fast-moving goods; where deliveries are made on a road network every day. We explore several computationally tractable allocation techniques that are good proxies for the Shapley value in problem instances of a size and complexity that is commercially relevant.
Researcher Affiliation Collaboration Haris Aziz EMAIL Data61/CSIRO and University of New South Wales (UNSW), Sydney, Australia Casey Cahan EMAIL University of Auckland, Auckland, New Zealand Charles Gretton EMAIL Hivery, Sydney, Australia; and Australian National University (ANU), Canberra, Australia; and Griffith University, Gold Coast, Australia
Pseudocode Yes Algorithm 1 DP-TSP-Shapley Algorithm 2 Appro Shapley Algorithm 3 Subset Shapley
Open Source Code Yes All code and data used in this project is available in a public Git repository at: https://github.com/nmattei/ Shapley TSG.
Open Datasets Yes Corpus available online at https://github.com/nmattei/Shapley TSG
Dataset Splits Yes For this tests we take the full Synthetic dataset and perform a 10-fold cross-validation (Bishop, 2006). To perform k-fold cross-validation, we take the dataset and break it into k equal sized folds F = {f1,..., fk}.
Hardware Specification Yes All timing experiments reported were performed on a computer with an Intel Xeon E5405 CPU running at 2.0 GHz with 4 GB of RAM running Debian 6.0 (build 2.6.32-5-amd64 Squeeze10).
Software Dependencies No The paper mentions using "Concorde (Applegate et al., 2007)", "Sci Kit Learn (Pedregosa et al., 2011)", and "Sci Py" but does not provide specific version numbers for these software components.
Experiment Setup Yes No. Samples = 300,000 2.812 (2 142)2 4 0.752 283,052 We choose to limit our linear model to two highest scoring elements φ DEPOTand φ MOATas these are both significantly higher scoring than the others and adding more elements may cause overfitting.