# datagents

**Repository Path**: macdows/datagents

## Basic Information

- **Project Name**: datagents
- **Description**: No description available
- **Primary Language**: Unknown
- **License**: Not specified
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2026-03-02
- **Last Updated**: 2026-03-11

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# AutoData-Agent

Multi-agent automated data analysis SaaS platform powered by LangChain and LangGraph.

## Features

- **Automated Data Cleaning**: Handle missing values, outliers, and type conversions
- **Statistical Analysis**: Descriptive statistics, correlation analysis, distribution analysis
- **Data Visualization**: Auto-generated charts (histograms, scatter plots, heatmaps)
- **Feature Engineering**: Standardization, encoding, PCA
- **Machine Learning**: Auto model selection, training, and evaluation
- **Report Generation**: Comprehensive markdown reports with insights

## Quick Start

### 1. Install Dependencies

```bash
# Using uv (recommended)
uv sync

# Or using pip
pip install -e .
```

### 2. Configure API Key

```bash
# Copy the example env file
cp .env.example .env

# Edit .env and add your OpenRouter API key
# OPENROUTER_API_KEY=your-api-key-here
```

Get your API key from [OpenRouter](https://openrouter.ai/).

### 3. Run the Application

```bash
streamlit run app/main.py
```

## Architecture

```
datagents/
├── config/           # Settings and prompts
├── app/              # Streamlit application
├── agents/           # 7 specialized AI agents
│   ├── orchestrator.py    # Supervisor agent
│   ├── cleaning.py        # Data cleaning
│   ├── eda.py             # Statistical analysis
│   ├── visualization.py   # Chart generation
│   ├── feature_engineering.py
│   ├── automl.py          # ML modeling
│   └── reporting.py       # Report generation
├── core/             # Core modules (LLM, executor, state, graph)
├── utils/            # Utility functions
├── outputs/          # Generated files (charts, models, reports)
└── temp/             # Temporary file storage
```

## Usage

1. Open the application in your browser
2. Enter your OpenRouter API key in the sidebar
3. Upload a CSV file (max 50MB)
4. (Optional) Enter your analysis requirements
5. Click "Start Analysis"
6. View the generated report, charts, and download results

## Agent Workflow

1. **Orchestrator**: Reads data schema, creates execution plan
2. **Cleaning**: Handles data quality issues
3. **EDA**: Performs statistical analysis
4. **Visualization**: Generates charts
5. **Feature Engineering**: Prepares data for ML
6. **AutoML**: Trains and evaluates models
7. **Reporting**: Aggregates results into final report

## Tech Stack

- **Frontend**: Streamlit
- **LLM**: OpenRouter API (DeepSeek)
- **Framework**: LangChain + LangGraph
- **Data**: pandas, numpy, scikit-learn
- **Visualization**: matplotlib, seaborn

## License

MIT