• R is a language and environment for statistical computing and graphics
• Supported by the R Foundation for Statistical Computing
• R language is based on the S Language, developed at Bell Labs by John Chambers in the 1970s.
  
• The 1988 version 3 rewrite of S is the basis for R as it exists today.
• Versioned releases of R go back to 2004, version 0.49.
• Runs on Windows, Mac, and Linux platforms

My short answer:

• As a statistical dabbler, I can't afford a SAS license.
• R is …
  • Free to use
  • Comparable in many respects to commercial software
  • Large and growing user base
  • Widely available instruction, tutorials, tips & tricks
• Practically everything touted as a reason for R could easily be said about other software packages.

• This is something of a religious question. My only experience with statistical software is using R and Excel. Excel is mostly useful for very simple situations.

• If you have built a career using SAS or another package, I would not discourage you from continuing down that road.

• If you prefer to have a wide assortment of tools at your disposal, then learn R. It will add to your personal and our organizational capacity. Plus it will build character!

• R is moderately difficult to learn
• Fortunately there are many resources for training available
• Coursera
• Lynda.com
• DataCamp
• There are also a wide range of resources for finding answers and tips
• RBloggers
• Rexercises
• Quick-R

My experience has been through the Coursera Data Science specialization. This is a 10 course series of month-long classes (I've completed 9 of 10) taught by professors from John Hopkins Bloomberg School of Public Health.

This doesn't make me an expert, just a slightly experienced dabbler.

• R by itself is a command-line based statistical engine.
• The R runtime download includes a very basic GUI, imaginatively named "RGUI"
• The native RGui is functional but lacks a lot of features you might expect from an full IDE, or integrated development environment.
• With RGui, you still are basically working with the command line interface.

• There are many alternative GUI tools for working with R
• The most widely used GUI tool is the RStudio IDE
• Rstudio operates as adjunct to native R, but adds useful conveniences

• CRAN is the Comprehensive R Archive Network
• Contains nearly 9,000 R packages
• Reasonable vetting of packages assures they will work, but no guarantees.
  
• CRAN Task views provide topical guidance for CRAN packages

• Other resources: BioConductor (for genomic data), GitHub (the Wild West!)

• GIT is a distributed version control system that can easily be integrated with RStudio
• Developed by Linus Torvalds, principal developer of Linux
• Designed for distributed, non-linear workflows
• Useful for iterative versioning of data management and analyis tasks
• Extremely useful for collaborative versioning, either publicly or privately, via GitHub or a private Git server
• Widely used, with many books and tutorials available

Most general purpose programming languages operate on single variables, with code littered with loops to process arrays of information

g <- rnorm(100000)
h <- rep(NA, 100000)

# Start the clock!
ptm <- proc.time()

# Loop through the vector, adding one
for (i in 1:100000){
  h[i] <- g[i] + 1
}
# Stop the clock
proc.time() - ptm

##    user  system elapsed 
##    0.19    0.00    0.18

h[1:10]

##  [1]  0.9159057  0.2007966  2.1333830 -0.1743988 -0.4212987  1.6411912
##  [7] -0.5610115  0.9808929  1.1687580  1.0009902

vectors in R are processed easily, without loops.

ptm <- proc.time()

# operations occur over the entire vector g
h <- g + 1

proc.time() - ptm

##    user  system elapsed 
##       0       0       0

h[1:10]

##  [1]  0.9159057  0.2007966  2.1333830 -0.1743988 -0.4212987  1.6411912
##  [7] -0.5610115  0.9808929  1.1687580  1.0009902

And quickly!

• Based on Leland Wilkinson's Grammar of Graphics book, GGPlot is a robust package for plotting graphics developed by Hadley Wickham and supported by RStudio

• The Big Idea: graphics have
  • data and aesthetic mappings (color, size, location)
  • geometric objects (bars, lines)
  • scales (linear, log, )
  • facet specification (different views of data)
• Bar graphs, line graphs, histograms, density plots, box plots, etc.

• Other graphics options: baseR, lattice, rgl, gdal, ggmap, plotly, D3

library(ggplot2)
set.seed(314159)
df <- data.frame(X=rnorm(1000,mean=50,sd=20),Y=rnorm(1000,mean=50,sd=10))
df$relation <- df$X <= df$Y
p <- ggplot(df,aes(x=X, y=Y)) +
      coord_cartesian() +
      scale_x_continuous() +
      scale_y_continuous() +
      scale_color_manual(name="X <= Y",values =c("TRUE"="red","FALSE"="blue")) +
      geom_point(aes(color=factor(relation))) +
      geom_smooth(method="lm",se=TRUE) + 
      labs(title="Using GGPLOT2", x="Random X", y="Random Y")

• Essential shortcuts for subsetting, summarizing, rearranging, and joining together datasets.
  • Filter - for filtering data
  • Mutate - add new calculated variables to a dataset
  • Sample - select random rows from a dataset
  • Summarise, Group_by, Arrange, Count, and more
• DPLYR has a useful pipe operator that makes it very intuitive to chaining DPLYR operations together

      airquality %>% 
          filter(Month != 5) %>% 
          group_by(Month) %>% 
          summarise(mean(Temp, na.rm = TRUE))

• lubridate - useful for working with dates
  • date format parsing
  • timespan aware (last month, three months, an hour and half )
  • time-zones
• foreign - useful for importing data from many different formats
  • Epi Info
  • Minitab
  • SAS
  • SPSS
  • Stata
  • …and more
• validate - test your data against pre-defined rules, either in or cross-data set
  • ex: "Height should be > 0", "height-weight ratio should be > 0.5"
  • Evaluates how many data items were checked against each rule
  • Evaluates how many items passed, failed or resulted in NA
  • Evaluates Whether the check resulted in an error (could not be performed) or gave an error