Details & weights

Select a node in the Tree to open the blade, then choose DETAILS.

Fields

Field	Meaning
Label 1 … N	The business-label path leading to the selected node.
Selected	The node you clicked.
Type	The available data type for the node.
Weight	The node’s share of all currently selected leaves.
# of Variables	Selected leaves under this node compared with all leaves beneath it.

Understand weight

Weight is a selection aid. It shows how much of the current field selection belongs to the node you are reviewing.

For example:

• If 4 leaves are selected and the current branch contains 1 of them, its weight is 25%.
• If the branch contains 3 of those leaves, its weight is 75%.
• An unselected leaf has no weight.

Weights update when you tick or untick leaves elsewhere in the Tree.

Use weight during curation

Review weight to answer:

• Does one source or branch account for most of the selection?
• Did I include more fields from one area than intended?
• Are the fields required for my use case represented?
• Can I remove fields that do not contribute to the intended questions?

Weight does not rate data quality or guarantee that a field is important. It only describes the current selection.

Review variables

The variable count helps you understand coverage inside the selected branch. For example, 3 of 10 means that three leaves are selected beneath a branch containing ten available leaves.

Use it with the label path and field names. A higher count is not automatically better; the appropriate selection depends on the intended Agent questions.

Before training

Before clicking Train Model:

1. Review the fields central to the task.
2. Check that unrelated fields are not selected.
3. Confirm that each source branch has the intended representation.
4. Review AI Data Advisor suggestions, if available.
5. Give the model a name that describes its purpose.

Continue to Semantic dataset.