feature-suggestion-agent 0.1.6

An intelligent agent for suggesting features for machine learning datasets

Feature Suggestion Agent

The Feature Suggestion Agent is an intelligent agent that uses LLMs (Large Language Models) to automatically generate and suggest features for machine learning tasks. It can produce feature definitions, formulas, reasoning, and SQL queries for dataset transformation.

Features

• Accepts dataset metadata, target information, and modeling approach.
• Generates new feature suggestions using an LLM.
• Provides:
  • Feature name (title)
  • Columns used (columns_involved)
  • Calculation logic (formula_logic)
  • Reasoning for usefulness
  • SQL to create the feature
  • Aggregation/grouping info for window or group operations
• Parses LLM responses into validated Pydantic models.

---

📦 Installation

Prerequisites

• Python 3.11+
• Git
• uv – A fast Python package and environment manager.
  • For a quick setup on macOS/Linux, you can use:

    curl -LsSf https://astral.sh/uv/install.sh | sh
    

Setup

1. Clone the repository

   git clone https://github.com/stepfnAI/feature_suggestion_agent.git
   cd feature_suggestion_agent/
   git checkout main
   

2. Set up the virtual environment and install dependencies This command creates a .venv folder in the current directory and installs all required packages.

   uv sync --extra dev
   source .venv/bin/activate
   

3. Set up environment variables

   # Optional: Configure LLM provider (default: openai)
   export LLM_PROVIDER="your_llm_provider"
   
   # Optional: Configure LLM model (default: gpt-4.1-mini)
   export LLM_MODEL="your_llm_model"
   
   # Required: Your LLM API key (Note: If LLM provider is opeani then 'export OPENAI_API_KEY', if it antropic 'export ANTROPIC_API_KEY', use this accordingly as per LLM provider )
   export OPENAI_API_KEY="your_llm_api_key"
   

🛠️ Usage

Basic Usage

python examples/basic_usage.py

🧪 Testing

Run the test file:

pytest -v tests/test_agent.py 

🏗️ Architecture

The Feature Suggestion Agent is built with a modular architecture

• Core Components:

  • agent.py: Base agent implementation
  • models.py: Data models and schemas
  • constants.py: prompts
  • config.py: model configurations

• Dependencies:

  • sfn-blueprint: Provides core framework functionalities and utilities for agent development.
  • pydantic: Data validation

🤝 Contributing

📝 License

[Add your license information here]

🤝 Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

📧 Contact

[e-mail]