Rabbit - Multi-Touch Attribution Platform

A full-stack platform for analyzing marketing touchpoint data using attribution models. Features a React frontend and FastAPI backend with confidence scoring and business insights.

Quick Start

Docker (Recommended)

git clone https://github.com/ChristopherLandaverde/rabbit.git
cd rabbit
docker-compose up --build

• Frontend: http://localhost:3000
• Backend API: http://localhost:8000

Manual Setup

# Backend
python3 -m venv venv && source venv/bin/activate
pip install -r requirements.txt
python3 run.py

# Frontend (new terminal)
cd frontend && npm install && npm run dev

Features

• 5 Attribution Models: Linear, Time Decay, First Touch, Last Touch, Position-Based
• Interactive Frontend: React dashboard with file upload, model selection, and results visualization
• State Persistence: Auto-save, session recovery, and analysis history
• Docker Support: One-command deployment for development and production
• File Support: CSV, JSON, and Parquet files
• Confidence Scoring: Statistical confidence for attribution results

Data Format

Your CSV file should include these columns:

Column	Required	Description
timestamp	✅	When the touchpoint occurred
channel	✅	Marketing channel (email, social, etc.)
event_type	✅	Type of event (click, conversion, etc.)
customer_id	❌	Customer identifier
conversion_value	❌	Value of conversion

Example:

timestamp,channel,event_type,customer_id,conversion_value
2024-01-01 10:00:00,email,click,cust_001,
2024-01-01 13:00:00,email,conversion,cust_001,100.00

API Usage

Upload and analyze data:

curl -X POST "http://localhost:8000/attribution/analyze" \
  -F "file=@data.csv" \
  -F "model_type=linear"

Response:

{
  "results": {
    "total_conversions": 3,
    "channel_attributions": {
      "email": {
        "credit": 0.4,
        "conversions": 1,
        "revenue": 90.0,
        "confidence": 0.85
      }
    }
  }
}

Attribution Models

Model	Description
linear	Equal credit to all touchpoints
time_decay	More credit to recent touchpoints
first_touch	All credit to first touchpoint
last_touch	All credit to last touchpoint
position_based	Weighted by position (40% first, 40% last, 20% middle)

Testing

# Run all tests
python3 scripts/run_tests.py

# Run specific test types
pytest tests/unit/          # Unit tests
pytest tests/integration/   # Integration tests
pytest tests/performance/   # Performance tests

Development

# Backend development
python3 run.py

# Frontend development
cd frontend && npm run dev

# Docker development
docker-compose -f docker-compose.yml -f docker-compose.override.yml up --build

Documentation

• State Management Guide
• Docker Setup Guide
• API Documentation

License

MIT License - see LICENSE file for details.