The Statistical Significance of "R"
- 1. What is R? What can you do with R? Getting Help Conclusions The Statistical Significance of R Premal P. Vora 1 1 Penn State Harrisburg School of Business Administration Middletown, PA 17057. fpv at psu.edu October 19, 2009 Premal P. Vora The Statistical Significance of R
- 2. What is R? What can you do with R? Getting Help Conclusions What is R? R is a system for statistical computation and graphics. It consists of a language plus a run-time environment with graphics, a debugger, access to certain system functions, and the ability to run programs stored in script files. Released under GNU GPL. Officially a part of GNU. OS: Linux, Unix, Windows and Mac CPUs: i386, alpha, arm, hppa, ia64, m68k, mips/mipsel, powerpc, s390, x86_64, powerpc-apple-darwin, mips-sgi-irix, i386-freebsd, rs6000-ibm-aix, and sparc-sun-solaris. Premal P. Vora The Statistical Significance of R
- 3. What is R? What can you do with R? Getting Help Conclusions Why should anyone care? “I keep saying that the sexy job in the next 10 years will be statiticians” said Hal Varian, chief economist at Google. “And I’m not kidding.” From a NY Times article published on August 6, 2009. Available at http://www.nytimes.com/2009/08/06/technology/06stats.html Drowning in data. Many, many closed commercial statistics packages available but not clear whether there is one winner. R has been widely and enthusiastically embraced in academia and in industry. R is open source, fast, solid, extensible. Premal P. Vora The Statistical Significance of R
- 4. What is R? What can you do with R? Getting Help Conclusions Origins of R Created by Ross Ihaka and Robert Gentleman both at the University of Auckland (New Zealand). Follows the language definition of S as much as possible. S created at Bell Labs (Becker, Chambers, Wilks). R has lexical scoping – S does not. A lot like Scheme (Steele and Sussman) under the hood. Now developed by the R Development Core Team. Current stable release 2.9.2 (released on 2009-08-04). Available at http://www.r-project.org Premal P. Vora The Statistical Significance of R
- 5. What is R? What can you do with R? Getting Help Conclusions Fundamental purpose of R A software tool for making inferences from data. Syntax and structure of language allows researcher to focus on asking the right questions and on coaxing answers from the data. Answers are trustworthy. Premal P. Vora The Statistical Significance of R
- 6. What is R? What can you do with R? Getting Help Conclusions Philosophy of R A language for expressions and for assignments. Expressions are evaluated and the result is immediately displayed. Assignments also evaluate an expression, but the result is assigned to an object and not printed. All assigned namespaces are held in memory. A facility to save and to load assigned namespaces is available. Premal P. Vora The Statistical Significance of R
- 7. What is R? What can you do with R? Getting Help Conclusions Simple numerical computation Operators: +, -, *, /, ^, ... Logical operators: ==, >, >=, !=, &, |, ... Hundreds of mathematical, statistical, and other functions: sqrt, log, log10, cos, tan, sum, min, max, mean, median, sort, ... Functions operate on numbers and a variety of data objects. Premal P. Vora The Statistical Significance of R
- 8. What is R? What can you do with R? Getting Help Conclusions (Data) objects Data objects: scalars, vectors, matrices, lists, dataframes. Objects can contain numbers, strings, logical quantities, or other objects. All elements in vectors and matrices must be of the same “mode” (R converts non-conforming elements on the fly when necessary). A list is a flexible data object that can contain other data objects, each of a different mode. A dataframe is like a matrix but each vector in that matrix is of a particular “mode”. Allows a collection of data of different modes to be treated as one object. Premal P. Vora The Statistical Significance of R
- 9. What is R? What can you do with R? Getting Help Conclusions Visualization/Graphics Base R automatically loads packages for creating visual displays of data. Very strong in this area. Graphics are customizable. Many open-source add-on packages for graphics are available. You can always write your own using the graphics primitives in R if dissatisfied with what’s available. Premal P. Vora The Statistical Significance of R
- 10. What is R? What can you do with R? Getting Help Conclusions Extension Language for creating your own functions. Collect a group of functions into a package and share it with others if you like. Get feedback from the user base. Premal P. Vora The Statistical Significance of R
- 11. What is R? What can you do with R? Getting Help Conclusions Live Demo Simple numerical computations. Math functions. Logical comparisons. Vectors, vector arithmetic. Reading a dataframe. Making tabular summaries of data. Visualization of x,y data. Running linear regression: y = a + bx + e. Premal P. Vora The Statistical Significance of R
- 12. What is R? What can you do with R? Getting Help Conclusions Help Sources Built-in help: questionmarkfunction Built-in help: help.search(“subject”) Built-in help: example(functionname) Online help: At the project website...several well-written manuals Online help: Several mailing lists including R-help Some mailing lists are devoted to specific special-interest groups such as R-SIG-Finance, R-SIG-ecology, etc. Support from third-party commercial firms is also available (for a fee). Premal P. Vora The Statistical Significance of R
- 13. What is R? What can you do with R? Getting Help Conclusions Strengths Fast, reliable, powerful, flexible, solid alternative to commercial packages. Well-known in academia and in industry. Many packages (2,000?) for different tasks available. Active development, eager support. Availability of support from third-party commercial firms makes it viable for proprietary use. Open source, GPL License. Premal P. Vora The Statistical Significance of R
- 14. What is R? What can you do with R? Getting Help Conclusions Weaknesses (All personal observations) Memory management restricts data set size to size of(RAM + swap space - space occupied by other processes). Learning curve for basic to intermediate usage is relatively flat, but thereafter steep. Premal P. Vora The Statistical Significance of R