> ML_DATASET // GERMAN-CREDIT-DATA_v1.0
Statlog (German Credit Data)
UCI Machine Learning Repository / Prof. Hofmann (Univ. Hamburg) · Banking & Credit Risk · 1,000 credit applications, 20 attributes
Banking & Credit RiskCC-BY-4.01,000 credit applications, 20 attributesopen
Dataset Profile & Characteristics
Label Type:Binary creditability (1: Good, 2: Bad credit risk)
Languages:de, en
License Tier:permissive-open-source
Modalities:tabular
Intended Use
- Cost-sensitive binary classification benchmark (cost matrix: false good = 5x penalty)
Prohibited / Discouraged Use
- Production lending credit scoring without modern regulatory validation
Bias, Leakage & Privacy Risk Analysis
Privacy / Sensitive Data Risks:
Simulated applicant financial and personal attributes (marital status, age, housing).
Known Bias:
1990s West German financial demographics; gender and marital status bias.
Known Benchmark Leakage:
Duration of credit and existing checking account balance have dominant correlation with default.