Skip to content

Featrix API for Agents

This documentation is optimized for AI agents and automated systems. It provides structured, comprehensive API information for programmatic access.

Quick Reference

from featrixsphere.api import FeatrixSphere

featrix = FeatrixSphere()

# Train a Foundational Model
fm = featrix.create_foundational_model(data_file="data.csv")
fm.wait_for_training()

# Create and train a predictor
predictor = fm.create_binary_classifier(target_column="target")
predictor.wait_for_training()

# Make predictions
result = predictor.predict({"col1": "value", "col2": 123})
print(result.predicted_class, result.confidence)

Contents

Quick Start Guides

Document Description
CHEATSHEET.md Minimal copy-paste patterns (start here)
SKILLS.md Complete task-based workflows
TROUBLESHOOTING.md Common issues and solutions

API Reference

Document Description
api-reference.md Complete API reference with all methods, parameters, and return types
MODEL_CARD_SPEC.md Model card specification and fields

Use Case Guides

Step-by-step implementations for common ML tasks:

Use Case Guide Best For
Churn Prediction Customer retention prediction Subscription businesses, SaaS
Fraud Detection Transaction fraud classification Financial services, e-commerce
Similarity Search Find similar records Recommendations, deduplication
Regression Numeric value prediction Pricing, demand forecasting

Key Concepts

Object Hierarchy

FeatrixSphere (client)
    |
    +-- FoundationalModel (learns data structure)
            |
            +-- Predictor (classifier or regressor)
            |       |
            |       +-- APIEndpoint (production serving)
            |       +-- PredictionResult --> PredictionFeedback
            |
            +-- VectorDatabase (similarity search)
            |
            +-- ReferenceRecord (positive-only matching)

Workflow Pattern

  1. Create Foundational Model - Train on your data (learns column relationships)
  2. Create Predictor - Train classifier or regressor on target column
  3. Make Predictions - Single or batch predictions with confidence scores
  4. Deploy - Create API endpoint for production use
  5. Monitor - Configure webhooks for alerts, send feedback for improvement

Installation

pip install featrixsphere

Support