AI-powered Marketing Mix Modeling assistant with natural language interface.
┌─────────────────────────────────────────────────────────────────┐
│ 📊 SAGE - AI MMM Copilot │
├─────────────────┬───────────────────────────────────────────────┤
│ SIDEBAR │ CHAT INTERFACE │
│ │ │
│ ⚙️ Configuration│ ┌─────────────────────────────────────┐ │
│ │ │ 👤 User │ │
│ 🔑 OpenAI Key │ │ "What is TV's ROI?" │ │
│ ✅ Loaded │ └─────────────────────────────────────┘ │
│ │ │
│ ─────────── │ ┌─────────────────────────────────────┐ │
│ │ │ 🤖 SAGE │ │
│ 📊 Data Source │ │ "TV has an ROI of 1.34x. This is │ │
│ ◉ Upload CSV │ │ within the industry benchmark of │ │
│ ○ Sample Data │ │ 0.8-1.5x..." │ │
│ │ │ │ │
│ ✅ Data loaded: │ │ [Interactive Plotly Chart] │ │
│ 260 rows │ │ ┌────────────────────────────┐ │ │
│ │ │ │ ROI by Channel │ │ │
│ Channels: │ │ │ ┌─TV──────────■ 1.34 │ │ │
│ TV, Search, │ │ │ ┌─Search──────■ 2.45 │ │ │
│ Social │ │ │ └─Social─────■ 1.89 │ │ │
│ │ │ └────────────────────────────┘ │ │
│ 📋 Preview Data │ └─────────────────────────────────────┘ │
│ [Expandable] │ │
│ │ ┌─────────────────────────────────────┐ │
│ ─────────── │ │ 💬 Ask me anything about your MMM │ │
│ │ │ data... │ │
│ ✅ Ready! │ └─────────────────────────────────────┘ │
└─────────────────┴───────────────────────────────────────────────┘
- 🤖 Natural language queries for MMM analysis
- 📈 Automatic response curve fitting (Hill curves)
- 💰 Budget optimization with constraints
- 🧠 Domain knowledge (RAG with ChromaDB)
- 📊 Interactive Plotly visualizations
- 🔄 Feedback system with alternative answers
- Go to streamlit.io/cloud
- Sign in with GitHub
- Click "New app"
- Select this repository
- Main file:
app.py - Deploy!
# Clone
git clone https://github.com/adityapt/SAGE.git
cd SAGE
# Install dependencies
pip install -r requirements.txt
# Run
streamlit run app.py- Enter OpenAI API Key (sidebar)
- Upload CSV or use sample data
- Ask questions!
Example queries:
- "What is TV's ROI?"
- "Allocate 100M across channels"
- "Show me response curves"
CSV with columns: date, channel, spend, impressions, predicted
See data/sample_template.csv for example.
MIT