I have an binary classification AutoML model that runs as versioning each month. The model pipeline ran and retrained overnight without issue. The new version of the model was produced (see image)
However, when I look at the confusion matrix that is being generated within Vertex, it has added a category "Dropped".
This category doesn't exist in the dataset that the model is trained on. Any ideas why this has occurred would be appreciated.
User | Count |
---|---|
2 | |
1 | |
1 | |
1 | |
1 |