Find best.

Parameters:

Parameters:
See dedicated page for more information.
The FindBest action returns the name or content of the column with the "best" value, based on user-defined criteria. This is especially useful when working with multi-class classification models where each class is scored in a separate column, and you need to determine the winning class.
| C1 | C2 | C3 | C4 |
|---|---|---|---|
| 14 | 5 | 643 | 320 |
List of columns to inspect:
Select the columns where the action will look for the best value.
Example: C1, C2, C3, C4
Number of top best values to search for:
Defines how many top values to retain. For most use cases, this is set to 1 to find the single best column.
Definition of best value:
Choose the criteria used to define the "best" value.
largest: Selects the column with the highest value.smallest: Selects the column with the lowest value.largest in absolute value: Selects the column with the largest magnitude, ignoring sign.smallest in absolute value: Selects the column with the smallest magnitude, ignoring sign.Output:
Defines what should be written in the result column.
column's names: The output will be the name of the column containing the best value (e.g., C3).content of "Label" columns: If label columns are specified, the result will contain their content corresponding to the best column.Optional "label" columns:
Allows you to define a set of label columns that align with the columns being inspected. When using the content of "Label" columns output mode, values from these columns will be returned.
Name of the column(s) with the result:
Specify the name of the output column that will store the result.
Example: Best
| C1 | C2 | C3 | C4 | Best |
|---|---|---|---|---|
| 14 | 5 | 643 | 320 | C3 |
See the following screenshot for an example:

NOTE
- When using
content of "Label" columns, make sure your label columns match in order and number with the columns you’re inspecting.- Use
largest in absolute valueif your data contains both positive and negative values and you only care about the magnitude.- This action is ideal when post-processing classification model outputs or rule-based decisions.
