You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Isn't PredictionFeature a subset of Feature so shouldn't there be some link between them?
And I have a more in-depth question:
All these are assumed to be columns in a dataset (I assume name refers to a column)
But a dataset may include multiple CSVs; if the same col name is used in several tables, how do you distinguish it?
Shouldn't you link up to CSVW for a more formal description of CSV columns? Eg it has a lot more specific description of col types than your "feature type" codelist
Furthermore, some datasets are not tabular. Eg can OML describe ML over semantic datasets?
As currently defined, OML has a number of related things that are not defined the same:
oml:Feature: incominghasFeature, targetFeature; attributesname, type, description`oml:PredictionFeature: incoming hasPredictionFeature; attributesname, typeoml:defaultTargetAttribute: comma-separated string don't use comma-separated multiple values #13oml:ignoreAttribute: stringsoml:rowIDAttribute: stringQuestions:
PredictionFeaturecannot have adescription?PredictionFeaturea subset ofFeatureso shouldn't there be some link between them?And I have a more in-depth question:
namerefers to a column)CSVWfor a more formal description of CSV columns? Eg it has a lot more specific description of col types than your "feature type" codelist