​
Model Management API

This document outlines the RESTful API endpoints for managing machine learning models in the Plexe platform.

​
Endpoints

​
Get All Models

GET /models

Retrieve a list of all models associated with your account.

Response:

[
  {
    "model_id": "model_20240327_001",
    "model_name": "Customer Churn Predictor",
    "status": "completed",
    "created_at": "2024-03-27T12:00:00Z",
    "intent": "Predict customer churn based on usage patterns",
    "provider": "auto",
    "metric_name": "accuracy",
    "metric_value": 0.87
  },
  {
    "model_id": "model_20240327_002",
    "model_name": "Product Recommendation",
    "status": "processing",
    "created_at": "2024-03-27T14:30:00Z",
    "intent": "Recommend products based on purchase history",
    "provider": "auto"
  }
]

​
Create a Model

POST /models/{model_name}

Start training a new model using your uploaded data.

Path Parameters:

  • model_name (required): Name for your model

Request Body:

{
  "upload_id": "550e8400-e29b-41d4-a716-446655440000",
  "intent": "Predict customer churn based on usage patterns",
  "metric": "accuracy"  // Optional
}

Response:

{
  "model_id": "model_20240327_001",
  "model_name": "Customer Churn Predictor",
  "status": "pending",
  "created_at": "2024-03-27T12:00:00Z",
  "intent": "Predict customer churn based on usage patterns",
  "provider": "auto",
  "upload_id": "550e8400-e29b-41d4-a716-446655440000"
}

​
Get Model Details

GET /models/{model_name}/{model_version}

Retrieve details for a specific model.

Path Parameters:

  • model_name (required): Name of the model
  • model_version (required): Version ID of the model

Response:

{
  "model_id": "model_20240327_001",
  "model_name": "Customer Churn Predictor",
  "status": "completed",
  "created_at": "2024-03-27T12:00:00Z",
  "intent": "Predict customer churn based on usage patterns",
  "provider": "auto",
  "metric_name": "accuracy",
  "metric_value": 0.87,
  "input_schema": {
    "properties": {
      "usage_days": {"type": "number"},
      "last_activity": {"type": "string", "format": "date"},
      "plan_type": {"type": "string"}
    }
  },
  "output_schema": {
    "properties": {
      "churn_probability": {"type": "number"},
      "risk_level": {"type": "string"}
    }
  }
}

​
Check Model Status

GET /models/{model_name}/{model_version}/status

Check the training status of your model.

Path Parameters:

  • model_name (required): Name of the model
  • model_version (required): Version ID of the model

Response:

{
  "status": "completed",  // possible values: "pending", "processing", "completed", "failed"
  "result": {
    "metric_name": "accuracy",
    "metric_value": 0.87  // Only present when status is "completed"
  },
  "error": null  // Only present when status is "failed"
}

​
Get Model Description

GET /models/{model_name}/{model_version}/describe

Get a description or documentation for a model.

Path Parameters:

  • model_name (required): Name of the model
  • model_version (required): Version ID of the model

Response:

{
  "description": "This model predicts customer churn probability based on usage patterns and account history.",
  "features": ["usage_days", "last_activity", "plan_type"],
  "target": "churn_probability",
  "performance": {
    "accuracy": 0.87,
    "precision": 0.82,
    "recall": 0.79
  }
}

​
Perform Inference

POST /models/{model_name}/{model_version}/infer

Use a trained model to make predictions.

Path Parameters:

  • model_name (required): Name of the model
  • model_version (required): Version ID of the model

Request Body:

{
  "usage_days": 45,
  "last_activity": "2024-03-20",
  "plan_type": "premium"
}

Response:

{
  "churn_probability": 0.23,
  "risk_level": "low"
}