Methods
Summary
The beauty of this data-driven project is that we will combine various datasets together to address many planning and operational issues for EAST system. The datasets are from three ways: 1. the gigabytes of taxi trip data in New York city provided by taxi and limousine commission; 2. the UberX and UberPool trip and demand data in New York city collected by one own developed tool; and 3. online survey on demand for EAST.
The discrete choice model, that is one type of common models on transportation mode choices, will be employed to specify the residents' preference on EAST in New York city. In addition, the qualitative analysis on EAST will be implemented through summarizing related reports and strategies.
For those practical problems on planning and operations, various scalable methods are introduced. For instance, the clustering methods that can process both spatial and temporal characteristics will be used to deploy pickup, charging, and parking locations of EAST; the graph theory and heuristic solution methods will be developed to solve the NP-hard problem of large scale ride matching; and the game theory and equilibrium are adopted to discuss suitable fleet size and distribution.
For those who want to review the applications of these modeling structure, you are welcome to check out the works done by the Interdisciplinary Transportation Modeling and Analytics Lab at Purdue. We have introduced these modelings on a series of data-driven transportation problems.
Challenges
Three main challenges are as follows:
- We can not collect Uber trip destination directly, however, can estimate it through observing Uber vehicle movements;
- Spread the online survey to New Yorkers and get considerable feedback, however, we can try to overcome by incentives to spread and complete this survey; and
- Improve scalability of modelings by introducing new trends in big data science.
Protocols
This project has not yet shared any protocols.