EuroProcure-10-ML: A Benchmark Dataset for Procurement Fraud Risk Detection in European Public Procurement

EuroProcure-10-ML is a benchmark dataset for machine learning-based procurement fraud-risk detection in European public procurement. The dataset contains 204,752 post-award procurement records collected from Tenders Electronic Daily (TED), the official procurement platform of the European Union, covering the period 2016–2025. The dataset spans multiple procurement domains, including healthcare, digital technology, construction and infrastructure, automotive, green energy, defense and security, and aerospace. Each record contains structured procurement information such as buyer details, procurement procedures, contract characteristics, CPV classifications, funding indicators, award information, bidder participation measures, and theory-informed fraud-risk indicators. To support reproducible research, the dataset includes engineered risk signals derived from procurement audit literature, including single-bid tenders, value overruns, repeated buyer–supplier relationships, unusually short submission periods, and non-competitive procurement procedures. These signals are aggregated into a weakly supervised fraud-risk framework and provided alongside the dataset. Data quality was prioritized throughout dataset construction. Missing values were preserved whenever information was unavailable in the original procurement notices. No synthetic records, fabricated observations, or artificial imputations were introduced. Users should interpret missing values as reflecting source-data availability rather than processing errors. The dataset is intended for research in fraud detection, anomaly detection, public procurement analytics, explainable artificial intelligence, risk assessment, and public-sector machine learning. Citation DOI: 10.5281/zenodo.21355728 Raw data sourced from TED (Tenders Electronic Daily), European Union. Enriched and processed by Urwa Binat Khalid.