Data drift
AIGP glossary · Last reviewed: · By Victor Humenhuk (AIGP certified)
Data drift - Input data's statistical properties change over time vs training data; performance degrades.
In the AIGP body of knowledge, Data drift comes up under Module 8: AI Governance Vocabulary.
Data drift in context
- Treat every occurrence as an incident, keep records in an ==AI registrar==, and know the ==five usual causes== - brittleness, lack of robustness, lack of quality data, insufficient testing, and model or data drift. (Incidents, consequences and accountability)
- ==Intent is the dividing line==: disinformation is deliberate, misinformation is not - and data poisoning is an attacker corrupting training data, unlike natural data drift. (Risks, security and harms)
Where Data drift is covered in the AIGP study notes
Related terms
- Training data
- Validation data
- Input data
- Ground truth
- Corpus
- Data quality
- Data provenance
- Synthetic data
- Variables
- Preprocessing
Test yourself on Data drift
Recognising a definition is not the same as applying it in an exam scenario. Work through the free AIGP practice questions, or read the full AIGP study guide - every study note is free.