Study finds digital terroir models falter under independent spatial tests

Research on geographical indications showed machine learning accuracy often drops 11.8% without spatial validation, raising doubts for regulators