A dark-themed, interactive Streamlit dashboard for venue-level feature engineering analysis across restaurant delivery channels (In-Store, UberEats, DoorDash, Self-Delivery).
This dashboard transforms raw revenue, cost, and order data into actionable engineered features across 5 analytical groups — giving operators a unified view of channel performance, cost efficiency, and growth potential across 24 venues.
- Python 3.8+
git clone https://github.com/your-username/engineered-features-dashboard.git
cd engineered-features-dashboard
pip install -r requirements.txt
streamlit run app.pystreamlit
plotly
pandas
numpy
scikit-learn
The dashboard is organised into 6 tabs:
| Tab | Description |
|---|---|
| 📡 Channel Revenue Ratios | Share of total revenue per delivery/dine-in channel with heatmap & stacked bar |
| 💸 Cost-to-Revenue | COGS, OPEX, and delivery cost efficiency with waterfall chart |
| 💰 Profit per Order | Channel-level profit efficiency — dollars earned per order fulfilled |
| 🔗 Interaction Terms | Compound commission × channel-share signals with correlation matrix |
| 📈 Growth-Adjusted Demand | Forward-looking demand scaled by venue growth factor |
| 🔍 Cross-Group Summary | Radar chart + full correlation heatmap across all engineered features |
Channel Revenue Ratios — InStore_RevRatio, UE_RevRatio, DD_RevRatio, SD_RevRatio
Cost Ratios — COGS_to_Rev, OPEX_to_Rev, TotalCostRate, DeliveryCost_SDRev
Profit per Order — ProfitPerOrder, InStore_ProfitPerOrder, UE_ProfitPerOrder, DD_ProfitPerOrder, SD_ProfitPerOrder
Interaction Terms — CommRate_x_UEshare, CommRate_x_DDshare, DeliveryCost_x_SDshare
Growth-Adjusted Demand — GrowthAdj_Orders, GrowthAdj_Revenue, GrowthAdj_UEOrders, GrowthAdj_SDOrders
- Venue Filter — multi-select to focus on specific venues
- Chart Type — toggle between Lines + Markers, Lines only, or Markers only
- Highlight Metric — surface a focus metric across all views
- Dark UI with custom CSS (
#07090fbackground, cyan/amber/rose accent palette) - Fonts: Sora (headings/body) + IBM Plex Mono (data/labels)
- All charts built with Plotly — fully interactive and hover-enabled
├── app.py # Main Streamlit application
├── requirements.txt # Python dependencies
└── README.md
MIT License — free to use and modify.