A data model formally represents the entities, attributes, relationships, and constraints represented by data. For an ML project, understanding and documenting it establishes what data is available and what results may be realistic. section 2.4
Turn data-model uncertainty into explicit pre-project work:
- Create a short ML-readiness project when access or understanding is inadequate. section 2.4
- Secure provisions for privacy, data retention, permitted use, security, and data handling before extracting or inspecting data. section 2.4
- Document the model and inspect the data’s real characteristics before dimensioning the ML project. section 2.4
- Add substantial time and funding contingency when nobody can explain what the data contains. section 2.4
