CartoBoost Neural Embedding Regressor

Use NeuralEmbeddingStandaloneRegressor when the learned ID embedding model is the artifact you want to train, score, save, and serve directly. This works best when the train and prediction populations share stable IDs such as entity IDs, route pairs, account buckets, item IDs, or recurring operational groups.

Use When

• The embedding model itself is the artifact under test.
• Stable ids recur in both training and prediction data.
• You want a direct model, not embedding features for another estimator.

Python Example

from cartoboost.neural import NeuralEmbeddingStandaloneRegressor
import numpy as np

entity_id = np.array([132, 161, 132, 236, 161, 236], dtype=np.uint64)
dense = np.array(
    [
        [1.0, 6.0],
        [2.5, 8.0],
        [1.2, 6.0],
        [3.1, 17.0],
        [2.7, 8.0],
        [3.3, 17.0],
    ],
    dtype=float,
)
target = np.array([2.7, 3.1, 2.8, 3.4, 3.2, 3.5])

model = NeuralEmbeddingStandaloneRegressor(dim=4, n_estimators=20, random_state=7)
model.fit(entity_id, target, dense=dense)

pred = model.predict(entity_id, dense=dense)
mae = model.score(entity_id, target, dense=dense)
model.save("neural-standalone.json")

Interactive Example

Neural embedding browser model

Runs embedding in the browser with the bundled CartoBoost Wasm model.

Ready to run in this page.

Inputs

• fit(ids, y, dense=None) trains one supervised embedding model.
• predict(ids, dense=None) returns one prediction per row.
• score(ids, y, dense=None) reports mean absolute error.
• save(path) and load(path) persist the complete standalone artifact.

ids is a one-dimensional unsigned integer array. When dense is provided, it must have the same row count as ids.

Validation

Report random, temporal, and cold-ID splits separately when they support different claims. Under cold-ID or cold-route holdouts, report fallback behavior explicitly because unseen IDs cannot recover learned ID-specific effects.

Limitations

• Learned ID effects do not generalize naturally to unseen identifiers.
• Embeddings can memorize frequent entities under random splits.
• Inspect temporal and cold-ID results before treating embeddings as structural signal.