Simulating Passenger Flow Data Challenge

Data Challenge on Simulating Passenger Flow - Alstom Project - 1st Place (done in my personal time): • Designed a simulation using Dijkstra's algorithm and Graph Neural Networks (GNNs) to model passenger flow in disrupted metro networks while maintaining environmental sustainability standards. • Focused on metro resilience, anticipating passenger flows to better adjust train frequencies and identifying critical inter-stations requiring special monitoring. • Developed a virtual metro simulation to: - Visualize the impact of service disruptions on passenger flows - Propose alternative routes to redirect passengers efficiently • Analyze network vulnerabilities and improve long-term resilience • Created a dataset for AI training, identifying bypassable inter-stations and quantifying the impact of their removal on passenger traffic. • Optimized code and simulation performance to minimize environmental impact while ensuring solution effectiveness. • Presented the solution in a competitive context, showcasing practical applicability and business relevance. 💡 Skills Gained: • Data Visualization & Analysis: Interpreting complex metro network data for decision-making • Data Science, AI & ML/DL: Building and training models to simulate real-world scenarios • Graph Algorithms & GNNs: Applying Dijkstra and GNNs to model passenger flow and network disruptions • Problem-Solving: Designing realistic alternatives for disrupted networks • Responsible Development: Estimating solution consumption and integrating environmental considerations • Business & Communication: Understanding company needs and presenting solutions effectively in a competitive setting