Analyzing, Visualizing Noordin Top Terror Network (Gephi) University of Minnesota Outreach (UCINET Net Draw) Tools of the Trade When pen and paper (or even Excel) just won’t do https://www.joe.org/joe/2011december/a9.php Infantry Training Task Network (iGraph) Infantry Training Task Network (ORA) US Power Grid (Pajek) CDT Maldonado, CDT Talty, MAJ Koban https://www.cise.ufl.edu/research/sparse/matrices/Pajek/USpowerGrid.html UNCLASSIFIED
Why? We may not need to use analytical packages and visualization for graph theory. What are some differences between graphs and networks? To do real work with networks, complex networks, dynamic networks, you need to get comfortable with software. Analysis Visualization…I picture is worth a 1000 words…or more. UNCLASSIFIED
You’ve Got MANY Options In fact, more than 50 from just the following two sites: University of Gronigen: https://www.gmw.rug.nl/~huisman/sna/software.html Good ol’ Wikipedia: https://en.wikipedia.org/wiki/Social_network_analysis_software We’re going to discuss the following: UCINET (NetDraw) Python (NetworkX) R (iGraph) ORA Mathematica Pajek Gephi UNCLASSIFIED
UCINET (NetDraw) You can get it here: https://sites.google.com/site/ucinetsoftware/home Lots of history (been around since 1987), widely used in academia Actively maintained (2-day workshop hosted by Borgatti in April) Additional information about NetDraw in Chapter 4 of http://www.faculty.ucr.edu/~hanneman/nettext/ Pro: Flexible data import Many metrics Supports more complex network analysis Compatible with many visualization platforms Con: Steeper learning curve Online help is “sophisticated.” Difficult to filter data being viewed NetDraw visualization relatively poor, cannot format output. UNCLASSIFIED
Pajek You can get it here: http://mrvar.fdv.uni-lj.si/pajek/ Lots of history (been around since 1996) No maintenance since May 2016 (active prior to that) Helpful book at http://mrvar.fdv.uni-lj.si/pajek/pajekman.pdf Offers both Pajek (good for large networks) and Pajek XXL (good for huge networks). Pro: Good online documentation Many metrics Supports more complex network analysis Con: Steeper learning curve Primarily for complex mappings of very large networks UNCLASSIFIED
Python (NetworkX) You can get it here: http://networkx.github.io/ If you like Python and networks, you’ll love NetworkX! Pro: Written completely in Python Good online documentation Many metrics Supports more complex network analysis Algorithms are easy to find and manipulate Con: May have poor performance and/or memory usage, depending on which algorithms are being called (do to dictionary data structure). Outputs are relatively simplistic and difficult to manipulate UNCLASSIFIED
R (iGraph) You can get it here: http://igraph.org/r/ … If you like R and networks, you’ll love iGraph! ## Download and install the package install.packages("igraph") ## Load package library(igraph) Pro: Many data analysis features Standard graph statistics are present Faster than NetworkX (>>) Con: API takes time to get used to (for Python users). Tools designed for analysts by analysts UNCLASSIFIED
ORA ORA: Organizational Risk Analyzer You can get it here: http://www.casos.cs.cmu.edu/projects/ora/download.php Or here: http://www.casos.cs.cmu.edu/projects/ora/ Used in TRADOC’s Advanced Network Analysis and Targeting (ANAT) course Pro: Very low barrier to entry. Dynamic and temporal network analysis Advanced network analysis Geo-spatial analysis tools. Con: Costs to use any functionality beyond ORA Lite ($480 per copy) Visualization is not as impressive as other platforms Difficult to modify existing algorithms UNCLASSIFIED
Mathematica It’s not just for Calculus! Pro: Tremendous amount of functionality with many available network and graph metrics. Student familiarity with the Mathematica interface. Con: Expensive to maintain license upon departure from USMA (loss of developed tools). UNCLASSIFIED
Gephi Get it here: https://gephi.org/ “Photoshop for graphs” Active developer community (last release FEB 2016) Very active online community (including Facebook users group) Pro: Premier graph visualization package Entirely GUI Intuitive to use Shallow learning curve Can visualize “Internet-sized” networks Con: Shorter list of analysis tools than other packages For very large graphs, must get “under the hood” to adjust memory usage settings UNCLASSIFIED
References http://www.slideshare.net/noahflower/overview-of-network- analysis-platforms http://pt.slideshare.net/keiono/overview-of-modern-graph- analysis-tools?smtNoRedir=1 http://files.meetup.com/1406240/sna_in_R.pdf UNCLASSIFIED