Model Training API

Endpoint

POST /api/model/train

Train a new decision model using labeled training data.

Authentication

Bearer JWT in Authorization header

Get authentication token

Request Parameters

Required Parameters

training_data

Array of labeled training examples

target_column

Target variable name

Optional Parameters

validation_split

Proportion for validation (default: 0.2)

early_stopping

Enable early stopping (default: true)

Example Request

POST /api/model/train HTTP/1.1
Authorization: Bearer <JWT_TOKEN>
Content-Type: application/json

{
  "training_data": [
    {
      "age": 34,
      "income": 85000,
      "decision": "approve"
    },
    {
      "age": 19,
      "income": 12000,
      "decision": "reject"
    }
  ],
  "target_column": "decision",
  "validation_split": 0.3,
  "early_stopping": true
}

Note: Replace <JWT_TOKEN> with your active authentication token

Response Format

{
  "status": "success",
  "model_id": "dec-ai-model-v1-3a2b98",
  "training_info": {
    "accuracy": 0.876,
    "precision": 0.842,
    "recall": 0.865,
    "f1_score": 0.853,
    "training_time": "3m 27s"
  },
  "validation_metrics": {
    "accuracy": 0.824,
    "precision": 0.798,
    "recall": 0.812
  },
  "links": {
    "model_insights": "/api/model/dec-ai-model-v1-3a2b98/analyze"
  }
}

Model will be automatically saved and available for prediction through model_id

Implementation Considerations

Training data must be under 50MB in size
Rate limits: 10 training requests per 24 hours
Models are deleted after 30 days of inactivity