Statistics and Data Science Seminar: An Interdisciplinary Approach for Identifying Partisan Gerrymandering

This seminar will be hosted jointly with the Political Science Department.

Abstract: Important insights into redistricting can be gained through an interdisciplinary approach that combines research from many fields, including statistics, operations research, computer science, high performance computing, math, law, and political science. We develop an Evolutionary Markov Chain Monte Carlo (EMCMC) algorithm for sampling maps that are conceptualized as spatial partitions that lie within a large, complex, and constrained spatial state space. The algorithm incorporates an optimization algorithm that adaptively updates a Markov chain. Our work integrates insights from all of these disciplines to create a novel approach for analyzing and reforming redistricting in a way that is tightly coupled with the framework that the Supreme Court has outlined over the past 5 decades.

Host: Robert Lunde