Spike: Define behavior for non-default woodwork logical types
- Langage dominant
- Python
- Étoiles
- 850
- Forks
- 96
- Métriques de merge des PR
- Aucune PR mergée en 30 j
Description
We have good coverage for what I would call the "default" woodwork types like Double, Boolean, Datetime, NaturalLanguage but how should we handle non-default types that represent more specific data types, e.g.
- Other logical types with string physical type -- Address, EmailAddress, Filepath, PersonFullName, PhoneNumber, URL
- Other categorical types with category semantic tag -- PostalCode, SubRegionCode, CountryCode
- LatLong
- Ordinal
- Timedelta
The outcome of this spike is a document detailing
- How we currently handle these logical types -where does `AutoMLSearch` fail and where does it not.
- A proposal for how to handle these types in `AutoMLSearch`, e.g. detect them and add specific components to handle them, modifying our existing components to handle them
Guide de contribution
Ouvrir le guide de contribution
Évaluation
Cette issue n'a pas encore été évaluée.