Introduction and Caveats
This page serves, as one might expect, to allow other researchers to access data analysis tools developed by the NCASD lab. Many of our tools can also be found on the Statnet Project site, or via the Comprehensive R Archive Network.
Caveats: This software is provided on an unsupported, as-is basis, under the terms of the GNU Public License. It is not guaranteed to be bug-free, let alone efficient; use it at your own risk. While the author is always happy to hear about bugs (and is sometimes happy to hear about suggestions and/or demands), he makes no promises whatsoever about making changes to the software, now or in the future. Amen.
Getting and Using R
The R statistical computing environment is distributed under the GNU Public License, and can be freely obtained either as source or as precompiled binaries for a wide range of platforms. The main R Project site has a great deal of information on the project and software, as well as pointers to other resources. For downloads, documentation, and the like, the Comprehensive R Archive Network (CRAN) site (or one of its mirrors) is the standard resource. Full details on installing and using the latest version of R can be found at these sites. Newcomers to R may also want to check out the various manuals and tutorials listed on the CRAN web site.
The Statnet Project
Most of the tools developed by the NCASD lab are part of the Statnet Project, a collaborative effort to develop Free Software tools for network analysis. Statnet incorporates packages such as sna and network, among many others, in a single interoperable toolkit. Find more information regarding Statnet at http://statnet.org. Statnet project packages with significant authorship by NCASD lab members include:
- ergm: Tools to support simulation and inference for exponential family random graph models.
- network: Core data structures and visualization tools for complex network data (including support for metadata, missing data, etc.).
- networkDynamic: Core data structures for dynamic network data.
- relevent: Tools for relational event models.
- sna: Tools for a range of social network analysis tasks, including classical descriptive analysis, null hypothesis tests, network regression, network visualization, data manipulation, and more.
Information on downloading and installing these and related packages can be found on the Statnet Project web site.
Spatial Data Analysis
US Census Suite The UScensus2010 suite of packages is comprised of spatial and demographic data for the 50 states and Washington DC at four different geographic levels (block, block group, tract, census designated place, and county) from the 2010 census. It also contains a number of functions for selecting and aggregating specific geographics or demographic information such as metropolitan statistical areas, counties, etc. The UScensus2000 suite contains similar data from the 2000 census. Files: UScensus package files, UScensus2010.pdf (Manual), and an article on the packages published in the Journal of Statistical Software.
The networkSpatial Package The networkSpatial package provides tools for simulation and analysis of spatially-embedded networks, including spatial Bernoulli graph generation, SIF inference, and more. Files: networkSpatial_1.00.tar.gz (Source).
Miscellaneous Older Packages (Not Necessarily Up to Date!)
The nettheory Package, v2.0The nettheory package is a collection of routines related to network generation and process theory, including models of power, propinquity, dominance, and diffusion. The primary function of the package ispedagogical, but it may have some research value as well. To use under UNIX, execute the command `R CMD INSTALL /mypath/nettheory_2.0.tar.gz’ from your command prompt. Under Windows, install the binary version(nettheory_2.0.zip) into your library directory (see your Rdocumentation). After the installation completes, use`library(nettheory)’ from within R to access the nettheory Library. Seethe R documentation for details. Note: nettheory requires sna v1.0 or higher.Files:Current INDEX andreference manual in Acrobat(PDF)format nettheory_2.0.tar.gz (Source) and nettheory_2.0.zip (Binary) package files, and nettheory-manual.2.0.pdf (Manual) [Current] nettheory_1.0.tar.gz (Source) and nettheory_1.0.zip (Binary) package files, and nettheory-manual.1.0.pdf (Manual)
The Metamatrix Package for Organizational Analysis, v0.1This package of R routines implements the metamatrix approach to therepresentation and analysis of organizational structure, as articulated byKathleen Carley, David Krackhardt, Yuquing Ren, and others. Thispackage builds on the SNA package (see above), and the latter must beinstalled to use many of the former’s routines. To use under UNIX, execute the commend `R CMD INSTALL /mypath/metamatrix_0.1.tar.gz’from your command prompt. Under Windows, unzip the package binary(metamatrix_0.1.zip) in your library directory (see the Rdocumentation). After installation completes, use`library(metamatrix)’ from within R to access the metamatrixlibrary. See the R documentation for details. Files:Current INDEX and reference manualin PostScript and Acrobat (PDF)formatsmetamatrix_0.1.tar.gz (Source)and metamatrix_0.1.zip (Binary) package files (Current)
Metamatrix Package Tutorial, v0.1This tutorial provides a quick and dirty introduction to the analysisof organizational structure using the Metamatrix v0.1 and SNA v0.3 packages. Be sure to get the associated data files, as these are needed to perform the exercises described in the text.Files: metamatrix.tutorial.0.1.pdf (Current) Tutorial data files(Current)
Yet Another Canonical Correlation Analysis Package, v1.0 The yacca (“Yet Another Canonical Correlation Analysis”) package is a small library of routines for (as the name implies) canonical correlation analysis (CCA). As compared with R’s built-in canonical correlation analysis routine (“cancor”), yacca offers three benefits: (1) it computes a large number of diagnostic and other quantities which are important for the use of CCA in practical settings; (2) it supports native visualization of CCA output (including helio plots); and (3) it scales its output in a more conventional (and, perhaps, intuitive) way than does cancor. Some might also say that it is better documented. To use under UNIX, execute the command `R CMD INSTALL/mypath/yacca_1.0.tar.gz’ from your command prompt. Under Windows, install the binary version (yacca_1.0.zip) into your library directory (see your R documentation). After the installation completes, use`library(yacca)’ from within R to access the yacca Library. See the R documentation for details. Files: Current reference manual in Acrobat (PDF) format yacca_1.0.tar.gz (Source) and yacca_1.0.zip (Binary) package files, and yacca-manual_1.0.pdf (Manual) [Current]