Programming with R for Reproducible Research

Course Synopsis

See Overview page. Eventually more details at: Course Catalogue Data)

Prerequisites

• Both parts of
  "Using R for Data Analysis" (lecture in fall semester),
  or similar knowledge of R on at least an *intermediate* (" > beginner") level
  
  • Slides of "Using R .." (2013)
  • Slides of "Using R .." (2012)  (with slightly more material in part 2)
  • The "textbook" of "Using R": Longhow Lam (2010). An Introduction to R freely available from CRAN.
• Laptop with R (>= 3.0.1) and one of RStudio / StatET / ESS, (or similar) "R IDE" installed.   Two students may team up, using one computer.
• One semester of (introduction to) statistics

Start of lectures

Tuesday, 23.02.2016

Lecture material

Week 1:

- Organization, Topics, etc: Emacs org (source), pdf.
- R code during browsing of "Using R.." chapter 7: 2014-02-18-ex.R.
- R markdown ("Rmd") file first.Rmd, conveniently opened in Rstudio, demonstrating both R markdown with its HTML, i.e. web content output. The resulting (first.Rmd --> first.md --> first.html) is currently available as public web page on Rpubs.
- "Everything (in R) is an object" --> explorations and a table [Rpubs] and its Rmd source

Week 2:

- Your questions about "Using R" (and the material above) : ? ...
- && vs &  and   || vs |    -- ?Logic
- coercion:     ?c ?Extract
- More on functions, notably closures:
  The 2 (and three) parts of a function: Commented R code and R markdown.
- Excursion: Exploring R packages and functions in there: Sweave from "Using R" or just the "one" slide
- Using the R code from Matloff's book.
  Our edition of original Ch7/envexample1.R. Using Rstudio's "compile as notebook" (*.R -> *.Rmd -> *.html, the latter with knitr) gives envexample1+ splinefun.

Week 3:

- Our edition of original Ch7/bookvec.R
- our example "text corpus" text1.txt
- Our version of original Ch4/findwords.R; and the (more efficient!)
split() version of original Ch6/findwords.R. Note seq_along(.)
- Our modified excerpt of H.Wickham's functional programming chapter.

Week 4:

- [continuing "functional programming" (week 3, above)]
- The initial R session, somewhat extended, of How R Searches and Finds Stuff
- Functions -> environments:
- ls(), get(), assign(), find(), ls.str(), new.env(), parent.env(),
globalenv(), emptyenv(), and the first two figures in How R Searches and Finds Stuff

Week 5:

- one_counter() example in "functional programming" (above) --- please study as homework, ask in class.
- "R is slow" etc:
-- "Premature optimization is the root of all evil", Donald Knuth
Rather: Test, test, and test again; using all.equal(target, current, tolerance ~= 10^-8) // Good R packages do -> sub directory './tests/'
-- typical issue about for() loop from Stackoverflow
--> functions system.time() and proc.time()
-- User Q about 'matrix vs. data.frame' on the R-help mailing list (March 2014). -- a first look at Bill Dunlap's solution + more, as Rmd script.
-- Our (modified) Rmd on "Performance" from Hadley Wickham's book chapter "Performance".

Week 6:

- Continuing "Performance" (see week 5): "Measure, don't guess" --> Using Rprof() and microbenchmark
- The 'matrix vs. data.frame' R-help example continued see above).
- R's byte compiler (-> require("compiler"); ?cmpfun), see in the above *.Rmd
- Performance <-> Copying of R objects: "Traching memory" memory-copying.R script.
- Start looking at R packages, source and "binary"; at first packages and namespace: env-namespace.R

Week 7:

- R packages, in source and "binary"; browsing Notes of "Package writing" course (Rnw and R files) (and the the "one" slide from week 2)
- package.skeleton()
- Packages and their Namespaces: Why are namespaces needed: Rmd whyNamespaces.Rmd and its html whyNamespaces.html.
- What happens when you call library(<pkg>) ?  -- week7-pkg-namesp.Rmd
- Understand more of How R Searches and Finds Stuff: Script week7-namespace-pkg.R

Extras:

- a small script to get all methods of a generic function, nicely in a list, hidden or not.
- from lapply() to parLapply(): R's builtin package 'require(parallel)'

- What happens when you call library(<pkg>): week7-pkg-namesp.Rmd

Lecture attestation (Testat):

In order to obtain the ECTS credit you
have to pass the exam -- answering some questions, and writing R code - in a *.Rmd (R Markdown file) at the end of the teaching block,
specifically on April 15.

Recommended Reading

• Norman Matloff (2011) The Art of R Programming - A tour of statistical software design.
  no starch press, San Francisco. on stock at Polybuchhandlung (CHF 42.-); see online for data, R code.
• Hadley Wickham (2013 ff) Advanced R, online
  more advanced than our course; partly focused on his own packages
• Suraj Gupta (March 29, 2012)  How R Searches and Finds Stuff, online;
  Tough read, but helpful with its nice illustrations. Do consider Duncan Murdoch's note  about it with minor caveats.

Miscellaneous on Programming (with R)

• "Literate Programming" by Donald Knuth
• "The Elements of Programming Style" by Kernighan and Plauger: Wikipedia, ,
  Quotes