Learning the Value of Teamwork to Form Efficient Teams

Authors: Ryan Beal, Narayan Changder, Timothy Norman, Sarvapali Ramchurn7063-7070

AAAI 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We apply our model to predict team performance and validate our approach using real-world team performance data from the 2018 FIFA World Cup. Our model is shown to better predict the real-world performance of teams by up to 46% compared to models that ignore inter-agent interactions.
Researcher Affiliation Academia 1School of Electronics and Computer Science, University of Southampton, Southampton, SO17 1BJ, United Kingdom {ryan.beal, t.j.norman, sdr1}@soton.ac.uk 2The National Institute of Technology, Durgapur, West Bengal 713209, India narayan.changder@gmail.com
Pseudocode No The paper presents mathematical formulations (MIPs) in equations 4 and 6, but it does not include explicitly labeled pseudocode or algorithm blocks.
Open Source Code No The paper does not provide any specific links to source code repositories or explicitly state that the code for the described methodology is publicly available.
Open Datasets Yes To evaluate our models we use a dataset collected from the 2018 FIFA World Cup.5 The dataset breaks down the 64 games from the tournament into an event-by-event analysis where each event gives different metrics including: event type (e.g., pass, shot, tackle etc.), the pitch coordinates of the event and the event outcome. This type of dataset is industry-leading in football and used by top professional teams. Thus, we believe that this is a good, real-world, dataset with the richness and challenge appropriate to rigorously assess the value of our model. 5All data provided by Stats Bomb www.statsbomb.com.
Dataset Splits Yes To learn the model weights, we use a 10-fold cross-validation approach, splitting the dataset randomly into 70% training and 30% test.
Hardware Specification No The paper does not provide any specific details about the hardware (e.g., GPU, CPU models, memory) used to run the experiments.
Software Dependencies No The paper mentions using a Logistic Regression algorithm and a Random Forest approach, and mixed-integer programming (MIP) techniques, but does not provide specific software names with version numbers for reproducibility (e.g., Python, PyTorch, CPLEX versions).
Experiment Setup Yes To learn the model weights, we use a 10-fold cross-validation approach, splitting the dataset randomly into 70% training and 30% test.