{
  "$schema": "https://aigovernanceengineer.com/schemas/dataset-card.v1.json",
  "dataset_id": "loan-apps-2019-2025",
  "version": "2026-04-30",
  "name": "Consumer loan applications with 12-month outcomes",
  "owner": "retail-credit-data-owner",
  "steward": "credit-data-engineering",
  "description": "Historical applications with repayment outcomes, to train and test the affordability model.",
  "composition": {
    "record_count": 412000,
    "unit": "one loan application with its 12-month repayment outcome",
    "features": [
      "declared income",
      "account inflows",
      "existing debt",
      "loan amount",
      "outcome"
    ],
    "time_range": "2019-01 to 2025-03"
  },
  "collection": {
    "method": "Extracted from the origination and servicing systems.",
    "sources": [
      "origination-db",
      "servicing-db"
    ]
  },
  "provenance": {
    "upstream": [
      "https://data.example.org/catalogue/origination-db",
      "https://data.example.org/catalogue/servicing-db"
    ],
    "lineage": "https://lineage.example.org/datasets/loan-apps-2019-2025/2026-04-30",
    "aibom_ref": "credit-afford-03 AIBOM, component loan-apps-2019-2025"
  },
  "personal_data": true,
  "lawful_basis": "legitimate_interests",
  "special_category": {
    "present": false
  },
  "licence": {
    "name": "Internal data, owned by the organisation",
    "allows_training": true
  },
  "representativeness": {
    "populations": [
      "applicants in ES and PT, ages 18 to 80"
    ],
    "known_gaps": [
      "self-employed applicants with under 12 months of history",
      "applicants over 75"
    ]
  },
  "quality_checks": [
    {
      "check": "missing income under 1%",
      "result": "pass",
      "run_at": "2026-05-02T08:00:00Z"
    },
    {
      "check": "label leakage scan",
      "result": "pass",
      "run_at": "2026-05-02T08:05:00Z"
    }
  ],
  "intended_uses": [
    "training and testing credit-afford-03"
  ],
  "prohibited_uses": [
    "marketing",
    "employee assessment"
  ],
  "splits": [
    "train 2019-2023",
    "validation 2024",
    "test 2025"
  ],
  "retention": {
    "until": "2031-04-30",
    "rule": "Snapshot deleted five years after the last model trained on it is retired."
  }
}