DATAGEN is an AI-driven multi-agent data analysis and research platform. It uses LangChain, OpenAI models, and LangGraph to automate hypothesis generation, data analysis, visualization, and report writing.
Collecting history — the radar snapshots this repo daily. The trend line appears after 3 days of data (1 so far).
What it is
DATAGEN is an AI-powered data analysis and research platform that utilizes multiple specialized agents to streamline tasks such as data analysis, visualization, and report generation. It leverages LangChain, OpenAI models, and LangGraph to handle complex research processes and integrates various AI architectures for performance.
How it works
The system employs a multi-agent architecture with agents for hypothesis generation, processing, visualization, code generation, literature/web search, reporting, quality review, and note-taking. It uses LangGraph to manage a state graph of the research workflow: 1) Hypothesis generation, 2) Human choice to continue or regenerate, 3) Processing (data analysis, visualization, search, report writing), 4) Quality review, 5) Revision as needed. Agents can be configured via agent_models.yaml to select provider and model_config per agent. The workflow supports dynamic task distribution and real-time coordination among agents. Advanced configuration includes a Progressive Disclosure approach and MCP integration for external services.
Getting started
- Clone the repository:
git clone https://github.com/starpig1129/DATAGEN.git
- Create and activate a Conda environment:
conda create -n datagen python=3.10
conda activate datagen
- Install dependencies:
pip install -r requirements.txt
- Rename and fill environment file:
# Rename .env Example to .env and fill all values
- Run the Python script:
python main.py
Recent releases
- No releases listed (Latest 0: none).
Traction
- Stars: 1779
License
- MIT
Getting started (config and usage details)
- Environment variables include WORKING_DIRECTORY, CONFIG_DIRECTORY, CONDA_ENV, CHROMEDRIVER_PATH, API keys for FIRECRAWL, CRW, OPENAI, ANTHROPIC, GOOGLE, LANGCHAIN, MCP, TAVILY, GITHUB_TOKEN.
- Agent configuration lives in CONFIG_DIRECTORY via agent_models.yaml, enabling provider and model_config per agent.
- Main components include hypothesis_agent, process_agent, visualization_agent, code_agent, searcher_agent, report_agent, quality_review_agent, note_agent.
Notes
- System requires API credits due to multiple API calls and may take time depending on task complexity.
- The system may modify analyzed data; backups are recommended.
- DATAGEN does not ship or train on provided datasets; users supply data at runtime via WORKING_DIRECTORY.