Conformal Online Auction Design

We provide the implementation codes in Python and present the corresponding results in Jupyter Notebook files.

Each experiment is contained in a separate folder. These folders include all relevant Python modules (.py files) and corresponding Jupyter Notebook files (.ipynb) for executing and reproducing the results. In addition, each experiment folder contains a dedicated README.md file that documents the purpose and usage of the associated modules and notebooks.

Folder Structure

• The "Real Data Analysis" folder contains the real-world eBay dataset obtained from https://www.modelingonlineauctions.com/datasets, and reproduces the analysis from Section 6 of our paper.

• The "Simulation Studies" folder includes four independent subfolders:

  • simulation_Section_7.1
  • simulation_Section_7.2
  • simulation_Supp_D.1
  • simulation_Supp_D.2

Each subfolder corresponds to a distinct experiment from Section 7 of the main paper or Section D of the Supplementary Material.

Reproducibility

The experiments are designed to be reproducible across platforms.

High-Performance Computing Environment

Experiments were primarily executed on a high-performance computing cluster under the following specifications:

• CPU architecture: Intel Gold
• Python version: 3.9.6
• Memory per core: 32 GB
• Execution interface: Jupyter Notebook

All required Python packages are listed in requirements.txt, generated via pip freeze. This ensures full reproducibility of our results on compatible systems.

Local Execution

All experiments can also be successfully executed on a personal computer. The notebooks have been tested using Python 3.10.13 on a Mac equipped with Apple Silicon (M4 Pro). A list of essential packages for local execution is provided in Mac_requirements.txt.