Basic network measures and visualization
Transcription
Basic network measures and visualization
5/11/16 Unit 3: Basic Visualization ICPSR University of Michigan, Ann Arbor Summer 2016 Instructor: Ann McCranie Compare three sociograms with exactly the same density…. N=50 N=100 N=500 1 5/11/16 Compare two more, same density, but one is undirected and the other is not…. Both N=50 “A process of charting has been devised by the sociometrists, the sociogram, which is more that merely a method of presentation. It is first of all a method of exploration. It makes possible the exploration of sociometric facts . . . It is at present the only available scheme which makes structural analysis of a community possible.” • J.L. Moreno. 1934. “Who Shall Survive? Foundations of Sociometry, Group Psychotherapy and Sociodrama.” Beacon, NY: Beacon House, Inc. (Quote from pp. 95-96 and via Lin Freeman’s work on visualization and social networks: http://moreno.ss.uci.edu/80.pdf 2 5/11/16 What makes a good network visualization? • It depends. • Do you want help “telling your story?” • Do you want to use visualization as a way to explore the structure of your networks? • Are you trying to highlight individuals or structure? • What are you trying to convey? This tells a powerful story… Bearman PS, Moody J, Stovel K. "Chains of affection: The structure of adolescent romantic and sexual networks" American Journal of Sociology, Vol. 100, No. 1. 3 5/11/16 Sometimes sociograms can truly be “maps” you can explore From the fantastic “Maps of Science” at http://mapofscience.com/ Sometimes they reveal subgroups J.K. Harris, K.E. Beatty, J.D. Lecy, J.M. Cyr, R.M. Shapiro II Mapping the multidisciplinary field of public health services and systems research Am J Prev Med, 41 (1) (2011), pp. 105–111 4 5/11/16 Things you must consider! • • • • Size of your network Nature of ties (directed? Weighted?) Density of your ties. What attribute information you need to show and what type of variables those are (categorical, continuous, etc) • Your publication/presentation outlet Basic elements for sociogram • Node – size, color, shape • Sometimes people color “rims” of nodes to denote additional information • • • • Labels – size, color, shape Ties – size, color and shape Arrows (showing directionality) Legend (for all those colors, sizes and shapes!) • Layout – hand-drawn, algorithm, etc. • Description to accompany it 5 5/11/16 Description to accompany your sociogram – what I recommend • • • • The data source The size of the network The density of ties The meaning of sizes, colors, and shapes (unless clear in a legend) • The layout – hand-drawn, layout algorithm • The software package you used The “spring embedder” and “forced-based layout” algorithms • Treat nodes and lines as if they are part of a spring-based or mechanical system. • These end up putting nodes with more shared ties closer together and moving others away • There are many solutions to how to accomplish this. • Popular ones are Fruchterman-Reingold, Kamada-Kawaii, and spring embedded. 6 5/11/16 Multidimensional Scaling Approaches • These are approaches that try to pair nodes closer in a two dimensional space together based on how “similar” they are to one another. • Each solution solves similarity in a different way. • Netdraw offers solutions based on: • Multidimensional scaling (in this case, they are using nonmetric MDS) • Principal components analysis • Gower Metric Scaling • You can then move nodes and lines around using other options (edge length bias and node repulsion, for instance) HOSPCONS, Netdraw, Clockwise from left: Gower metric scaling, MDS, PC, circle with optimization, spring embedding with geodesic distances + node repulsion + equal edge length bias, random 7 5/11/16 Suggestions for better visualizations… • Simpler is usually better. Instead of cramming a lot of information into one image, create a panel of different visualizations using the same layout so you can compare. • As you teach yourself how to look at graphs, you might want to pick a standard layout algorithm that you use so you become accustomed to it. More suggestions… • I often make my ties a lighter gray so they do not overwhelm my image • Anytime you have an undirected relation, DROP THE ARROWS. • Sometimes I only label key actors in large networks • Use software (Pajek, Visone, Network Workbench, statnet) where you can create SVG files and edit them. 8 5/11/16 Figure 1: Largest connected component of network of Recovery researchers, from ISI Web of Science citation data, 1991-2008, N=891. Note: The colors represent different CONCOR groups discovered in an initial analysis. Only key actors are labeled. The size of the nodes reflects the number of articles published during that time period. Layout is Fruchterman-Reingold in a square space in Pajek, which results in the characteristic round shape. Further modification was in image software to prepare for publication. 9 5/11/16 Comparisons • Perhaps you have multiplex relations, or you have networks that evolve over time. For print, you can choose panels. You can also produce dynamic network visualization. • See: • • “Dynamic Network Visualization: Methods for Meaning with Longitudinal Network Movies” American Journal of Sociology 110:1206-1241 SoNIA: http://www.stanford.edu/group/sonia/ Here follows a network that changes over time Co-authorship of Recovery Researchers from 1991-2008. N=891. Only key actors are labeled. The size of the nodes reflects the number of articles published during that time period. Layout is FruchtermanReingold in a square space in Pajek, which results in the characteristic round shape. Further modification was in image software to prepare for publication. 10 5/11/16 1991-1995 1996-2000 11 5/11/16 2001-2005 2006-2008 12 5/11/16 1991-2008 Modularity in the genetic diseasephenotype network FEBS Letters, Volume 582, Issue 17, Pages 2549-2554 Xingpeng Jiang, Bing Liu, Jiefeng Jiang, Huizhi Zhao, Ming Fan, Jing Zhang, Zhenjie Fan, Tianzi Jiang 13 5/11/16 Hidalgo, Klinger, Barabasi, Hausmann. 2007. http://scimaps.org/maps/map/ the_product_space_5/ It's a bird, it's a mosquito, it's a pig! The Spread of Pandemic Flu 14 5/11/16 http://well-formed.eigenfactor.org/index.html 15 5/11/16 Exemplary visualizations and resources • • • • • Butts, Carter T. (2009). ``Revisiting the Foundations of Network Analysis.'' Science, 325, 414-416. Kwang-Il Goh, Michael E. Cusick, David Valle, Barton Childs, Marc Vidal,and Albert-László Barabási. The human disease network PNAS 2007 104 (21) 8685-8690 Pablo Kaluza, Andrea Kölzsch, Michael T. Gastner, and Bernd Blasius. The complex network of global cargo ship movements J. R. Soc. Interface July 6, 2010 7:1093-1103 W. Van den Broeck, C. Gioannini, B. Goncalves, M. Quaggiotto, V. Colizza, A. Vespignani The GLEaMviz computational tool, a publicly available software to explore realistic epidemic spreading scenarios at the global scale. BMC Infectious Diseases, 11 :37 (2011) Krempel, Lothar. Network Visualization. 2011 The SAGE Handbook of Network of Social Network Analysis • Chapter 7 – Borgatti, Everett, and Johnson (2013) 16