Accra Transit Optimizer

🚀 Inspiration Accra faces growing urban mobility challenges — inefficient routes, overcrowding, unpredictable demand, and rising emissions. Public transport remains under-optimized due to a lack of data-driven planning tools. Inspired by these challenges and the availability of GTFS (General Transit Feed Specification) data, we set out to build an AI-powered system that empowers transport planners to forecast demand, optimize routes, and reduce CO₂ emissions while improving the commuting experience.

⚙️ What it does Accra Transport AI is an all-in-one dashboard and backend pipeline that enables:

📍 Route Visualization – Interactive maps to explore GTFS route, stop, and schedule data.

📊 Demand Forecasting – ML-based time series predictions of passenger demand.

🧠 Clustering – K-Means clustering of stops to identify route coverage issues.

🛣️ Route Optimization – Capacity-aware route optimization using real constraints.

🌍 Emissions Estimation – Route-wise, daily, and vehicle-type CO₂ footprint analysis.

🔄 Scenario Simulation – Impact analysis when a stop is blocked (e.g., roadworks, emergencies).

📁 Downloadable Reports – Exportable CSVs and emissions PDF summaries.

🖥️ Streamlit UI – Clean, dark-themed, mobile-optimized dashboard for planners and stakeholders.

🛠️ How we built it Frontend: Streamlit for interactive UI and visualizations.

Backend & ML:

Prophet for demand forecasting.

scikit-learn for clustering.

Custom heuristics for optimization.

Pandas-based GTFS preprocessing pipeline.

Data: GTFS data for Accra (stops, trips, routes, stop_times).

PDF Generation: Matplotlib + ReportLab to render emissions reports.

Visualization: Folium, Plotly, and Pandas plotting.

APIs: Modular design to expose REST endpoints for future integration.

🧱 Challenges we ran into 🔍 Parsing incomplete or inconsistent GTFS data (e.g., missing arrival_time, no shape data).

⏱️ Aligning timezones, schedule inconsistencies, and dealing with NULL timestamps.

📉 Forecasting low-frequency routes with sparse data.

🧮 Emission factor estimation due to lack of exact vehicle specs.

📦 Packaging everything into a single unified dashboard while maintaining performance.

🏆 Accomplishments that we're proud of Built a fully modular, end-to-end AI transit tool in under a week.

Integrated real transport data with advanced ML and optimization.

Delivered a dark-themed, mobile-friendly UI that supports planners and civic authorities.

Generated actionable insights like emission hotspots and block stop impact simulations.

Enabled report generation (PDF/CSV) to support offline use and policymaker briefings.

📚 What we learned How to clean and engineer features from GTFS data effectively.

The value of integrating data science, geospatial analysis, and optimization in urban planning.

How to use Streamlit and custom theming for professional-grade dashboards.

Handling edge cases in real-world data pipelines with fallback strategies.

🔮 What's next for Accra Transit Optimizer 🌐 Live GPS integration from vehicles to enable real-time tracking and incident alerts.

🧠 LLM integration (e.g., ChatGPT or Gemini) for a smart city transport assistant.

🚦 Traffic-aware route optimization using Google Maps or OpenStreetMap APIs.

📱 Mobile app version for passengers and city officials.

🌿 Sustainability reports with per-route CO₂ comparisons and green-route suggestions.

🛰️ Satellite + AI vision for road blockage detection and congestion prediction.

Built With

• fastapi
• python
• streamlit