`hxl2arff`: Attribute-Relation File Format (ARFF), focused for compatibility with WEKA, "The workbench for machine learning"
fititnt opened this issue · 1 comments
fititnt commented
- Attribute-Relation File Format (ARFF) specification
- File:
- Vocabulary/Taxonomy
- ID:
EticaAI-Data_HXL-Data-Science-file-formats
- Public URL: https://docs.google.com/spreadsheets/d/1vFkBSharAEg5g5K2u_iDLCBvpWWPqpzC1hcL6QpFNZY/edit#gid=1297379331
- ID:
TODO: add more information
fititnt commented
Different from hxl2tab
#2, the ARFF uses the 'class' (classifier) directly as would be an format type, and not as an extra attribute. Also, ARFF does not have an 'ignore' or 'skip' field direct in the file (if have to be done, I think is via interface).
But an very important point: Weka complains a lot if the exported file is not on a very strict format. It means that fields likely to be 'meta' or in special 'ignore' for Orange may not be exported as default for Weka. This may be pertinent if each exported field value is not validated to remove all characters that could make the Weka complaint.