Collaborative Research: A Whole-Community Effort to Understand Biases and Uncertainties in Using Emerging Big Data for Mobility Analysis

点击翻译按钮获取中文摘要

This NSF grant will quantify the biases and uncertainties associated with human mobility patterns when they are derived from big mobile data such as cell phone data, mobile app data and social media data. Information on human mobility patterns, or where and how Americans live, work and go about their daily activities, is the basis of hundreds of billions' investment for the nation's transportation infrastructures. These investment decisions have a direct impact on Americans’ upward social mobility, health, and well-being. The project is motivated by two factors: first, big mobile data increasingly replaces traditional household survey data in mobility analysis; and second, big mobile data is fundamentally unrepresentative (and biased) and a direct application of the results derived from such data can have substantial negative impacts on Americans’ health, prosperity and welfare. Novel education and outreach activities organically integrated with the research, including a collaboration with the Boston Museum of Science for a digital exhibit on mobility tales around the world, and a mini-track competition with MetroLab on “future mobility and justice for students around the world.”In addition to quantifying the biases and uncertainties associated with mobility patterns, this grant will also identify the extent those biases and uncertainties are affected by a number of factors, e.g., data characteristics, the modeling techniques used, and geographical differences. More specifically, the project comprises three research thrusts. Thrust 1 engages stakeholders and the research community to develop a solicitation calling for mobility labs around the world to submit critical mobility metrics, using their own data and methods. Thrust 2 involves the development of two novel methodologies: a coupled Bootstrap computational framework to quantify biases and uncertainties associated with derived mobility metrics and a rule-based learning framework to handle sparsity issues that likely arise during the analysis stage. Thrust 3 involves all participating labs for results summarization and dissemination. The project will unite multiple disciplines from transportation engineering to systems engineering, computer/information science, and social science in a concerted effort for better understanding the uncertainties and biases in mobility analysis when big mobile data is used. The results from the project will also provide practical insights for practitioners in using big mobile data for mobility analysis.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.