apart

Search Shortcut cmd + k | ctrl + k

Documentation

apart

Downloads 59this week

GitHub stars 0

Extension descriptor (YAML)

Evaluate affine decision trees in SQL

Maintainer(s): jokasimr

Installing and Loading

INSTALL apart FROM community;
LOAD apart;

Example

LOAD apart;

-- Split on x + y = 1, then on x - y = 0.
SELECT x, y,
       decision_tree(
         {
             weights: [
                 [1.0, 1.0],
                 [1.0, -1.0],
                 [1.0, -1.0]
             ],
             thresholds: [1.0, 0.0, 0.0],
             children: [[2, 3], [-1, -2], [-4, -3]],
             values: ['A', 'B', 'C', 'D']
         },
         x, y
       ) AS region
FROM (VALUES (1, 0), (0, 1), (-1, 0), (0, -1)) AS points(x, y);

About apart

apart evaluates affine decision trees. Each node compares a weighted sum of the inputs with a threshold, and each leaf supplies a result.

The tree stays constant for the query while the inputs vary by row. Inputs can be separate columns or a fixed-size array. Leaf values can use any DuckDB type, including lists and structs.

See the documentation and example use cases.