Institute of Transportation Engineers – 2010, 145 pagesThe R statistical package can best be thought of, in terms of user interface, as falling some-where in-between SAS and Limdep. Like SAS, R accomplishes different tasks using different procedures. Instead of the procedures being self-contained (like PROC REG in SAS), commands are issued one at a time, like Limdep. However, unlike SAS and Limdep, R can easily be extended using user-generated add-on packages, or libraries. Because R is open-source (and thus free), it is relatively easy to create add-on packages and quickly distribute them to a large, well-established community of users. Because of this, R can perform procedures and estimate statistical models that have not yet been implemented in purchaseable software such as SAS, Limdep, SPSS, and Stata. One primary example is spatial econometrics; while Stata contains capabilities for estimating certain types of spatial econometric models, the ma- jority of the most recent types of models (spatial panel models, spatial Probit, etc.) are only available in R. In short, R can give users access to cutting-edge methods and models easily.
The available packages from the central R website, the Comprehensive R Archive Network (CRAN, http://cran.r-project.org), can accomplish s variety of tasks, such as statistical analysis (which is to be expected from a statistical software package), machine learning, importing and manipulating GIS data (R's GIS capabilities are far stronger than SAS), parallel computing, qualitative research analysis, text mining, graph/network analysis, optimization, solving ordinary differential equations, image analysis, 3D graphics (through OpenGL), interactive graphics, and Bayesian inference. R can interface with a large number of other programs and platforms, such as Excel (through an add-in accessible through Excel), Google Maps, C++, WinBUGS, gretl, SAS, and even audio devices, relational databases (such as Oracle and SQL), the US Census database, LP_SOLVE, openNLP, PowerPoint, Word, LaTeX, the Web (through CURL), CPLEX, and Java. Despite this wide array of abilities, it is relatively easy to use a single data set with a wide variety of varieties, as will be demonstrated here.
One important noteunlike SAS, all commands in R are case-sensitive. While package names and commands in the text of the handout are denoted using all caps, the exact capitali-zations should be used as shown in the syntax examples.
