Authors: Zachary Neal*, Michigan State University, Rachel Domagalski, Michigan State University, Bruce Sagan, Michigan State University
Topics: Quantitative Methods, Urban Geography, Economic Geography
Keywords: network, spatial, backbone, bipartite, method
Session Type: Paper
Start / End Time: 9:50 AM / 10:05 AM
Room: Virtual Track 11
Presentation File: No File Uploaded
Spatial networks are often weighted, with the edges encoded with information about the frequency, intensity, or volume of exchanges between places. For example, transportation networks often encode the number of people moving from point A to point B, while global economic networks often encode the volume of potential financial transactions between two cities or countries. Although this information is important, because a weighted network it difficult to analyze and visualize, it is sometimes helpful to focus on its "backbone," which is a simpler binary network that contains only the most significant edges. This presentation will review a range of methods for extracting the backbone of spatial networks in a statistically principled way, so that the data reduction minimizes the loss of information. It will demonstrate these methods using the new BACKBONE package for R, focusing on two general contexts: (a) weighted spatial networks that capture flows such as transportation, and (b) weighted spatial networks that arise as bipartite projections.