DataInt, DataCategorical, DataBool - revise domain updates
Based on the example below we need to check and revise how the Options
domains are updated (for performance attributes based on the dataset samples):
- values seem to be of type
float
instead ofint
- for
DataInt
, the updated domain should contain all values in the min-max range. Now some values are omitted if they didn't appear in the dataset. - for
DataCategorical
, if the user specified options list (e.g.['cat', 'dog', 'bird']
), then probably this should not be updated to the dataset either - for
DataBool
the domain does not have to be updated, always use[True, False]
as default
>>> dataset.perf_attributes.dobj_list_orig
DataInt(name="win", dim=1, domain=Options([0, 1, 2, 3, 4, ...,995, 996, 997, 998, 999])
>>> dataset.perf_attributes.dobj_list
DataInt(name="win", dim=1, domain=Options([0.0, 1.0, 2.0, 3.0, 4.0, ...,167.0, 169.0, 170.0, 172.0, 182.0]), position=0, position_index=0)
Edited by Alessandro Maissen