Train Model

Create and train a new model.

POST /models/train

Train Model

cURL

curl --request POST \
  --url https://api.woodwide.ai/models/train \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: application/json' \
  --data '
{
  "model_name": "<string>",
  "model_description": "<string>",
  "dataset_version_id": "3c90c3cc-0d44-4b50-8888-8dd25736052a",
  "dataset_id": "3c90c3cc-0d44-4b50-8888-8dd25736052a",
  "label_column": "<string>",
  "input_columns": [
    "<string>"
  ]
}
'

202

{
  "model": {
    "id": "b2c3d4e5-0000-0000-0000-000000000001",
    "version_id": "b2c3d4e5-0000-0000-0000-000000000002",
    "version_number": 1
  },
  "job_id": "b2c3d4e5-0000-0000-0000-000000000003",
  "status": "queued",
  "woodwide_runtime_version": "0.1.0"
}

Reference training data by dataset_id (uses the latest ready version) or by a specific dataset_version_id for reproducibility.

Poll GET /models/{model_id} until status becomes ready (or poll GET /jobs/{job_id} for more detail).

See Capabilities for details on each model type, required parameters, and training metrics. See Output Formats for output columns by model type.

Authorizations

Authorization · string · header · required

Bearer authentication header of the form Bearer <token>, where <token> is your auth token.

Body

application/json

Request to train a new model.

Reference training data by dataset_id (resolves to the latest ready version automatically) or by dataset_version_id for reproducibility. In most cases, just pass dataset_id.

model_name · string · required

model_type · enum<string> · required

Canonical model types supported by the platform.

• prediction -- Supervised prediction (classification or regression) on a target column.
• anomaly -- Unsupervised anomaly detection. Returns anomaly scores and anomalous row IDs.
• embedding -- Generate dense vector embeddings for each row.
• clustering -- Unsupervised clustering. Returns cluster labels and descriptions.
• factors -- Factor analysis / dimensionality reduction.
• search -- Semantic nearest-neighbor search over the training dataset.

Available options: prediction, anomaly, embedding, clustering, factors, search

model_description · string

dataset_version_id · string<uuid>

Pin a specific dataset version. If omitted, the latest ready version is used.

dataset_id · string<uuid>

Dataset to train on (uses latest ready version). Provide this OR dataset_version_id.

label_column · string

Target column for prediction models. Required for prediction, forbidden for other types.

input_columns · string[]

Columns to use as features. If omitted, all columns except label are used.

Response

202

application/json

Successful Response

Response for training a model.

model · ModelVersionRef · object · required

Model version reference.

Show child attributes

job_id · string<uuid> · required

status · string · required

woodwide_runtime_version · string · required