Data Availability and Reproducibility

Clear documentation of datasets, computational methods, and experimental settings supports independent evaluation of research in data science and artificial intelligence.

Research data description

Authors are encouraged to describe the provenance, collection procedure, scope, version, and relevant permissions for the datasets used in their studies. Where data are publicly accessible, provide a stable repository link or identifier; where access is restricted, explain the restriction without exposing private or confidential information.

Method and computational transparency

• Document preprocessing and feature construction sufficiently to clarify what transformations were made.
• Report sample selection, training/validation/test splits, random seeds when material, and measures against data leakage.
• Describe compared methods, hyperparameter selection, baselines, evaluation metrics, and statistical procedures.
• Identify material software libraries, model versions, and computing environments when they affect reproducibility.
• Provide sufficient detail about model training and inference so that the results can be meaningfully assessed.

Code and model artifacts

Where permitted, authors may include links to source code, trained models, configuration files, or supplementary methods. Proprietary software, contractual obligations, and dataset licenses may limit disclosure; those limitations should be stated explicitly.

Suggested data availability statement

Responsible sharing

Public release must not compromise personal privacy, informed consent, intellectual property, confidentiality commitments, or legal protections. The ability to replicate an experiment is an important goal, but it does not supersede these obligations.