OVERVIEW

In This Presentation...

• Introduce the programming language and the IDE

• Discuss the advantages and disadvantages of R

• Outline the basic structure of R

• Show how to interact with the R environment

• Detail a few help resources

WHAT IS R?

In a Nutshell, R is...

• An open-source, interpreted computer language and suite of statistical operators for calculations on vectors and matrices

• A coherent, and integrated collection of tools and graphical facilities for data analysis and reproducible research

• A well-developed programming language which includes conditionals, loops, user-defined recursive functions, and input-output facilities

• One of the fastest growing technical programming languages in the world

IEEE Ranking Of Top Programming Languages - 2015 (click the image)

Where Did R Come From?

• The S programming language

  • Computing language originally developed at Bell Labs (1976) for data analysis

  • Licensed by AT&T/Lucent to Insightful Corp. under the product name: S-Plus.

• The R programming language

  • Written and released as open source software by Ross Ihaka and Robert Gentleman at University of Auckland

  • R plays on name "S"

  • Since 1997: international R-core team (~15 people) and thousands of code writers and statisticians share their work via R packages

WHY SHOULD I USE R? - HUMAN EFFICIENCY

Human efficiency vs. Machine efficiency

• Most programming languages are designed for Machine efficiency

  • Machine efficiency is concerned with producing numerical results as fast as possible

  • Pre-compiled languages (C, C++, FORTRAN) can produce numerical results extremely fast

  • However, pre-compiled code can be difficult to read for non-developers

  • Compiled languages were developed to make it easier to collaborate

  • Several compiled languages are proprietary - making it hard to connect with outside resources

• R was built for human efficiency

  • Human efficiency is concerned with how fast a language can produce a final product

  • As a compiled, interpreted language R may not compute as fast as pre-compiled languages

  • R might cost you minutes to compute a specific result, but...

  • R can save you days in getting your results compiled into a paper or presentation

  • Plus, R makes it easy to speed up your analysis by pulling in code from many other languages

Example of Human Efficiency with R

Fuel Efficiency Example

• Suppose we're interested in estimating which automobile design factors have a statistically significant effect on fuel efficiency

  • This type of analysis is often done using a technique called statistical linear regression

  • Right now, the how's and why's of linear regression aren't important

  • The focus of this example is how quickly R allows the user to go from a data set to a polished document showing the results of an analysis

• This example uses the Motor Trend Car Road Tests data set that installs with R - to view this data set type mtcars

  • Many R packages come with data sets pre-loaded, to view the names of the data sets type data(package = "datasets")

  • A new RStudio tab should open showing the names of the data sets in the datasets package

  • To read the documentation on a data set, type ?mtcars or help(mtcars)

Think About what you want...

• You may need to present your results to someone...what might they want to see?

  • The data set itself

  • Plots to illustrate and support your conclusions

  • A table of results from your analysis

• R/RStudio simplifies the process by bringing everything together

  • Data

  • Analysis (code, plots & results)

  • Presentation & Visualization

Data Set in Interactive Table

Plots to Support Conclusions (with code for better reproducibility)

Table of results from your analysis

\[ \begin{array}{rrrrr} \hline & Estimate & Std. Error & t value & Pr(>|t|) \\ \hline (Intercept) & 12.30 & 18.72 & 0.66 & 0.52 \\ cyl & -0.11 & 1.05 & -0.11 & 0.92 \\ disp & 0.01 & 0.02 & 0.75 & 0.46 \\ hp & -0.02 & 0.02 & -0.99 & 0.33 \\ drat & 0.79 & 1.64 & 0.48 & 0.64 \\ wt & -3.72 & 1.89 & -1.96 & 0.06 \\ qsec & 0.82 & 0.73 & 1.12 & 0.27 \\ vs & 0.32 & 2.10 & 0.15 & 0.88 \\ am & 2.52 & 2.06 & 1.23 & 0.23 \\ gear & 0.66 & 1.49 & 0.44 & 0.67 \\ carb & -0.20 & 0.83 & -0.24 & 0.81 \\ \hline \end{array} \]

ADVANTAGES AND DISADVANATAGES OF USING R

Advantages

• Only open-source language with a standardized set of development tools to extend the environment

• Gives full access to the programming environment, easily see what functions are doing, fix bugs, and extend software

• Promotes reproducible research by providing open and accessible tools for data analysis and reporting

• Provides a forum allowing researchers to explore and expand the methods used to analyze data

• Scientists around the world are the co-owners to the software tools needed to carry out research

• Over 7,000 R packages available (as of: 8/2015)

• The product of thousands of experts in many fields - R is CUTTING EDGE

Disadvantages

• Many percieve the R learning curve to be steep, minimal GUI
  Addressed by RStudio

  • In reality, R's learning curve is no steeper than that of any other language

  • The real issue relates to R how is structured compared to other languages

  • This is explained in more detail below

• Little commercial support; new users complain of hostility on R help sites

• Figuring out correct methods to use a function on your own can be frustrating

• Working with large datasets is limited by RAM
  Addressed by datatable package

• Data prep can be messier and more mistake prone in R vs. SPSS or SAS
  Addressed by dplyr and tidyr pacakges

How R is different

• In MatLab & JMP (for example) updates are released every 6-12 months

  • In preparing these updates, effort is made to ensure that the user experience is consistent across every function

  • This means that there is essentially a single learning curve to becoming proficient at using MatLab or JMP

  • A potential drawback of this development model is that users may not have access to newer methods until the the next update is released

• R is more like the wild west - updates and new capabilities are released every day

  • Some effort is made by the core R team to ensure a consistent user experience

  • But, R is purposefully focused on letting researchers from around the world share their work with the world quickly

  • A potential drawback of this development model is that it's NOT uncommon for multiple packages to have functions with the exact same name that produce completely different results

  • Thus, R does not have a steeper learning curve, but CAN, in some cases, have multiple learning curves for different packages

UNDERSTANDING THE R ENVIRONMENT

Basic Structure of R

• R was initially a functional programming language which has taken on attributes of an object-oriented programming during its development

• To understand computations in R, remember these five things

  • 1. Everything that exists is an object...almost
  • 2. Everything that happens is a function call
  • 3. Function call means that a function-class object "calls" a particular method
  • 4. Every object and function call has an associated environment
  • 5. Lexical scoping rules define how objects and functions can interact

Analogy for Understanding R's Basic Structure

• In cooking, simple methods are applied to combine ingredients into finished products

• At the most basic level these methods are:

  • Heating stuff up

  • Cooling stuff down

  • Breaking stuff apart

  • Mixing stuff together

• Specialized tools can adjust how a method is applied to certain "classes" of ingredients

  • Double boiler pan

  • Waffle maker

  • Microwave

  • Bread maker

• The process of scripting with R is similar to cooking

• At the most basic level some of the methods used in R scripting are:

  • plot - basic method for plotting simple numeric objects

  • print - basic method for printing simple numeric objects

  • summary - basic method for summarizing simple numeric objects

  • There are several others

• Specialized tools (functions) adjust how a method is applied to certain "classes" of objects

  • plot.function( ) - applies the plot method to function objects

  • plot.stepfun( ) - applies the plot method to step-function objects

  • plot.ts( ) - applies the plot method to time-series objects

  • You don't have to remember all the specialty functions - R picks the right one for you

Let's Examine What Happens When The Following Code Is Run

my.table <- xtable::xtable(mtcars)
print(my.table)

• Line 1 - xtable( ) first checks class(mtcars) = data.frame

  • The xtable( ) then searches for a method to convert a data.frame-class object to an xtable-class object

  • In this case, the function used is xtable.data.frame( )

  • R then stores the converted object under the name my.table

• Line 2 - print( ) then checks class(my.table) = xtable, data.frame

  • print then searches for a method to print an xtable-class object

  • In this case, the function used is print.xtable( )

R Packages

• Many R Packages create new, complex objects with new classes

  • The Survival package creates Surv-class objects

  • The xtable package creates xtable-class objects

  • The SMRD pacakge creats life.data-class objects

• Along with the new classes of objects, packages also define new methods for manipulating these objects

• THIS ABILITY TO CREATE NEW CLASSES OF OBJECTS IS THE POWER OF R

• Downloading packages expands our "kitchen", providing more sophisticated tools for creating and manipulating complex objects, SO...

INTERACTING WITH THE R ENVIRONMENT

Everything in R is an OBJECT...Almost

• Functions, vectors, datasets, character strings... are stored as objects

• Graphics are written out and are NOT stored as objects

• Objects are classified by two criteria:

  • MODE: how objects are stored in R - character, numeric, logical, factor, list, & function

  • CLASS: how objects are treated by functions - vector,list, matrix, array, data frame & hundreds of other classes created by specific functions

• Scripting is the process of matching objects and function calls such that the desired output objects (e.g., statistical results) and graphics are returned.

The Process of R Scripting - Step 1

Assigning names to objects - R is CasE sENsItiVE!!

• R has three methods to assign names to objects

  • Left assignment

  • Right assigment

  • Equals sign

• R treats each of these methods the same (in most cases)

• Object naming rules

  • Variable names can only contain letters and numbers separated by "." or "_"

  • Variable Names CANNOT begin with a number

var <- 5  ### Left assignment

200 -> Var ### Right assignment

meaning.of_life = 42 ### Or equal sign

var; Var; meaning.of_life

[1] 5 [1] 200 [1] 42

HELPFUL FUNCTIONS

Functions to access R object properties

• R scripting errors result when an object's properties don't match what was expected by the function call

• These functions are helpful to find what properties are assigned to an object

mode(object.name)  ### Returns an object's mode 

class(object.name) ### Returns an object's class

attributes(object.name) ### Returns the attributes associated with an object

str(object.name) ### Returns the complete list of properties assigned to an object

Functions to manage active objects

• Sometimes errors result if you accidentally overwrite an object in the current environment

• These functions are helpful for managing objects defined in the current working environment

ls( ) ### Returns a list of active objects in the current working environment

rm( ) ### Removes an object from the current working environment

rm(list = ls()) ### Removes all objects from the current working environment (use carefully)

Functions to manage your local file structure

• Errors can result if a script, saved in one file location, calls a file in another location

• These functions are helpful for managing your local file structure

• Additionally, creating projects in RStudio can help avoid these errors

getwd( ) ### Returns the location of the current working directory

setwd("C:/Users/Desktop") ### Reassigns the location of the working directory

file.choose( ) ### Opens a new file explorer window to choose a file

Help Resources Within R

?cos ### Searches the local package library for "cos"

??xyz ### Searches the full R documentation for "xyz" - also help(xyz)

vignette() ### Lists "how-to" demos available for each package in the library

help.search("t.test") ### Provides a categorized search of R documentation for "t.test"

Online Resources for Help

• Quick R http://www.statmethods.net/

  • Expands upon book: R in Action 2nd ed., Robert I. Kabacoff August, 2011, Mannings Publication Company, ISBN 9781935182399

• R-bloggers http://www.r-bloggers.com/

  • An aggregation of content collected from bloggers who write about R to help R users connect and follow the "R blogosphere"

• StackOverflow http://stackoverflow.com/questions/tagged/r/

  • A question and answer site for professional and enthusiast programmers

NOW YOU KNOW ENOUGH TO BE DANGEROUS

• Next, go beyond the basics and use R and RStudio to create and manipulate objects

• Check out these presentations

  • R: Environments and Objects

  • R: Plotting

  • R: Special Functions