Data Quality

The seed workflow starts from scrubbed multi-source outputs built by:

python scripts/build_complete_database.py
python scripts/seed_mongodb_collections.py --replace

Current local build summary:

• Normalized source rows: 40,309
• Unique deduped records: 8,173
• Duplicate source rows: 32,136
• Year range: 1500 to 2026
• Unique countries: 213
• Invalid fatal values: 0
• Records with coordinates: 1,196
• Invalid coordinate values: 0

Known limitations:

• Some historical records have missing country, year, or location fields.
• Species labels are source-derived and often descriptive rather than taxonomic.
• Incident counts do not equal risk because exposure data is not included.
• Kaggle import still requires authenticated Kaggle access.