Dynamic network evolution in ridesharing Abdulaziz Alhassan Shakiba Enayati Francis Niestemski Sabine Durand
Companies : Traditional Method: Taxis are run by liscenced taxi companies with specific taxi Ride Share Method: Anyone with can use their personal car and pick up customers
Map of Manhattan
The Model Nodes are placed on a grid Nodes are connected to neighboring nodes with a certain probability (smaller probability for diagonal edges) Customers and cars are placed randomly on the network Customers are assigned to the car that is closest to them Cars pick up customers and drop them off at their destination Customers disappear after being dropped off Cars remain at the drop off location until a new customer nearby requests a pick-up
Number of Customers Waiting Time _ _ Customers Cars
Waiting Time Edge Density Coefficient _ _ Customers Cars
_ _ Customers Cars Waiting Time Number of Nodes
Location Problem: Relocating the position of idle cars Routing Problem: Finding the best route for time window that each car will be available Set Covering Problem: Finding the ideal average number of cars in the system at each time unit
Random set of customers and cars joining the system gradually at each specific number of steps Having some priority rules for customers and modify the dispatching rule accordingly Considering time window of availability of cars in the system
Maximizing cars utility Minimizing the waiting time of cars and/or customers