Segment Transforms

Segment Transforms are Resources that map segments to numerical values. The primary purpose of Segment Transforms is to do lookalike modeling via index scores.

Under the hood, Segment Transforms are stored with a config. This is a dictionary where keys are segment GUIDs. The values are themselves are also dictionaries, where the keys are attribute names such as index_score, and the values are floats.

Creating Segment Transforms By Reference

When creating a segment transform by reference, a list of lists is passed along in a refs parameter (see Segment Transforms in the API for clarity). Each element of the refs parameter is a list, and each element of the sublists are integers that are segment transform IDs. The sublists are interpreted as cascading OR statements, while each list is ANDed together. So for example, the reference:

{
  "refs": [
    [1, 2, 3]
    [4, 5, 6]
  ]
}

corresponds to the logical statement: (1 OR 2 OR 3) AND (4 OR 5 OR 6).

Note that this is only available for certain types of segment transforms (i.e. those that have the underlying implementation of the combination procedure defined).

❗️

No Console Support

Segment Transforms are not support in the Console.