How dope-otaku/geospatial-demand-forecaster works
Geo-Demand AI simulates a Mumbai delivery fleet, streams rider coordinates through Kafka, clusters recent positions in a FastAPI service, and displays the live stream and demand visualizations in a web dashboard. The producer creates stateful riders; the backend consumes events and periodically attaches K-Means centroids before forwarding events over WebSockets. The dashboard tracks fleet metrics and renders rider, centroid, and hexagon layers with hover details. Backend and frontend relationships are supported by the excerpts; broader operational details are described in the README.
Main entry points: main.py Stream Service · page.tsx Live Dashboard · producer.py Rider Simulator
Components 7
Fleet Simulation
- Rider Simulator
data-pipeline/producer.py
Event Streaming
- Kafka Topic
- Stream Service
backend/main.py
Demand Analytics
- Demand Predictor
backend/main.py
Live Dashboard
- Live Dashboard
frontend/geospatial-demand-forecaster/src/app/page.tsx
Other components
- Map Tiles
- Dashboard Viewer
Connections 6
- Rider Simulator → Kafka Topic: publishes events
data-pipeline/producer.py - Kafka Topic → Stream Service: consumes events
backend/main.py - Stream Service → Demand Predictor: updates and queries
backend/main.py - Stream Service → Live Dashboard: streams WebSocket data
frontend/geospatial-demand-forecaster/src/app/page.tsx - Live Dashboard → Map Tiles: loads map style
frontend/geospatial-demand-forecaster/src/app/page.tsx - Dashboard Viewer → Live Dashboard: views