Helge Marahrens: Link Prediction Models for Globalization Research

Affiliation: Keough School of Global Affairs, University of Notre Dame
 

Abstract: The rapid expansion of data availability and computational methods has created enormous opportunities for social science research, particularly in the study of globalization. At the same time, these developments expose a fundamental tension between two modeling cultures: one centered on prediction and performance, and the other on explanation and theoretical understanding. This talk engages this divide by examining how link prediction methods from data science can be applied to long-standing questions in globalization studies. Focusing on the evolution of the worldwide firm-city network, I use link prediction models to analyze how connections between cities have formed and transformed over the past three decades (1993-2020). This approach captures globalization as a dynamic, subnational process, revealing some of the network generating mechanisms and speaking directly to theories of globalization and uneven development. By combining predictive modeling with substantive interpretation, the talk highlights both the promise and the challenges of integrating data science techniques into social science inquiry.

 

Bio: Helge Marahrens is an assistant professor of computational social science at the Keough School of Global Affairs at the University of Notre Dame and also an affiliate of Notre Dame's Lucy Family Institute for Data & Society. Marahrens's work intersects social science, data science and statistics. His research leverages new data sources to answer old questions and employs new methods to gain insights from existing data. Marahrens earned a Ph.D. in sociology (2023) and an M.S. in applied statistics (2019) from Indiana University. In 2021, he worked at Facebook’s core data science group and from 2023 to 2025, he worked as a postdoctoral fellow at Georgetown University's Massive Data Institute.