Anomaly: Added support for >1 target category columns #536
Conversation
oracle-contributor-agreement bot added the OCA Verified 
govarsha changed the title 
| 📌 Cov diff with main: 📌 Overall coverage: |
govarsha changed the title | 📌 Cov diff with main: 📌 Overall coverage: |
| for group, df in self.data.groupby(spec.target_category_columns[0]) | ||
| # Merge target category columns | ||
| | ||
| self.data["__Series__"] = utils._merge_category_columns(self.data, spec.target_category_columns) |
There was a problem hiding this comment.
It looks like the "Series" column gets added to self.data and then is referenced throughout the rest of the code.
Could you add to the docstring of this method "Merges all target category columns into a single column named "Series"" Or something equivalent.
I also see that the same code is run on line 180 of base_model.py. Do we need to do this twice?
There was a problem hiding this comment.
I will add the docstring.
Here _load_data in anomaly_dataset doesn't load/preprocess validation data. We are loading the validation data in base_model and the merge_category_columns is done there on validation data.
| if data.empty: | ||
| return total_metrics, summary_metrics, None | ||
| | ||
| data["__Series__"] = utils._merge_category_columns(data, self.spec.target_category_columns) |
There was a problem hiding this comment.
What is the intended behaviour when there is no target_category_column present?
There was a problem hiding this comment.
The method _load_data in anomaly_dataset doesn't load/preprocess validation data. We are loading the validation data here and the merge_category_columns is done here for that.
There was a problem hiding this comment.
And yeah I forgot to handle when no target_category_column is present for validation data. I will do that change
2 participants
ODSC-52180
Right now anomaly operator supports only one target category column. Added code to support more than 1.
Screenshots: