Intro to R

• Statistics
• Graphics
• Dynamic report generation
• Slide Shows
• Interface with the web
• Apps
• More

model <- lm(mpg~hp, data=mtcars)
summary(model)

Call:
lm(formula = mpg ~ hp, data = mtcars)

Residuals:
    Min      1Q  Median      3Q     Max 
-5.7121 -2.1122 -0.8854  1.5819  8.2360 

Coefficients:
            Estimate Std. Error t value Pr(>|t|)    
(Intercept) 30.09886    1.63392  18.421  < 2e-16 ***
hp          -0.06823    0.01012  -6.742 1.79e-07 ***
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Residual standard error: 3.863 on 30 degrees of freedom
Multiple R-squared:  0.6024,    Adjusted R-squared:  0.5892 
F-statistic: 45.46 on 1 and 30 DF,  p-value: 1.788e-07

• Switch between plain text and code
• Compile the report to a PDF, docx, html, or just about anything else
• More than one person working on the report? Use Subversion for version control!

A quick look at my blog for an example.

This slideshow was made in R!

Making slideshows is very similar to making dynamic reports.

library(RCurl)
library(data.table)
myfile <- getURL("http://statistics.cos.ucf.edu/mjohnson/wp-content/uploads/2013/08/CH06PR09.txt", ssl.verifyhost=F, ssl.verifypeer=F)
mydat <- fread(myfile)
summary(mydat)

       V1             V2               V3              V4        
 Min.   :3998   Min.   :211944   Min.   :4.610   Min.   :0.0000  
 1st Qu.:4193   1st Qu.:268759   1st Qu.:6.805   1st Qu.:0.0000  
 Median :4316   Median :291271   Median :7.325   Median :0.0000  
 Mean   :4363   Mean   :302693   Mean   :7.371   Mean   :0.1154  
 3rd Qu.:4472   3rd Qu.:321906   3rd Qu.:7.938   3rd Qu.:0.0000  
 Max.   :5045   Max.   :472476   Max.   :9.650   Max.   :1.0000  

The Shiny package allows easy app creation.

Example

See CRAN Task Views to see what R packages are available for your next project.

Some basic concepts of programming:

• Objects
• Logical operators && conditional logic (if/else)
• Vectors
• Indexing
• Functions
• Loops

We can create an object, x, via assignment.

x <- 5 

Then, when we ask for the value of x:

x

[1] 5

Logical operaters return boolean values.

x > 3

[1] TRUE

Conditional logic - a fancy term for if/else

if(x > 3){print('Yusssss')}

[1] "Yusssss"

x is a scalar. Objects can be lots of things, including vectors.

y <- c(1:10) # `c` is for concatenate
y

 [1]  1  2  3  4  5  6  7  8  9 10

R is particularly adept at operating on vectors.

y > 5 # operates on each element of the vector

 [1] FALSE FALSE FALSE FALSE FALSE  TRUE  TRUE  TRUE  TRUE  TRUE

Indexing allows us to retrieve individual elements from a vector.

y[1] # returns the first element of y

[1] 1

y[y > 5] # returns all elements of y where y > 5

[1]  6  7  8  9 10

Akin to macros in SAS, functions take input and return output.

sum(y) # sum() is a function

[1] 55

There is a function for just about everything in R. The sample() function randomly samples from a vector.

sample(y, 2) # sample from `y` two times

[1] 5 1

for(i in 1:5){
  print(i)
}

[1] 1
[1] 2
[1] 3
[1] 4
[1] 5

The Monty Hall Problem

Allow Kevin Spacey to explain.

(Famously solved by Marilyn vos Savant in her Parade magazine coulmn)

"There are only two hard things in Computer Science: cache invalidation and naming things." 
- Phil Karlton

doors <- c(1:3)
correctDoor <- sample(doors, 1)
guess1 <- sample(doors, 1)

Let's peak:

correctDoor; guess1

[1] 1

[1] 2

(Recall: correct door = 1, first guess = 2)

if(guess1==correctDoor){
  result <- "Winner!"
}else{
  result <- "Sorry :("
}
result

[1] "Sorry :("

• The host must reveal what is behind one door after your first guess.
• He cannot reveal the prize.
• If your first guess is wrong, he must reveal the only remaining incorrect door.

if(guess1!=correctDoor){
  doorToReveal <- doors[(doors!=correctDoor & doors!=guess1)]
}

(Recall: correct door = 1, first guess = 2)

If your first guess is correct, the host may reveal either of the two remaining doors.

if(guess1==correctDoor){
  canReveal <- doors[doors!=guess1]
  doorToReveal <- sample(canReveal, 1)
}

And the host reveals…

doorToReveal

[1] 3

(Recall: correct door = 1, first guess = 2)

Finally, you switch from your first guess to whichever remaining door that has not been revealed

remaining <- doors[(doors!=guess1 & doors!=doorToReveal)]
guess2 <- doors[remaining]
if(guess2==correctDoor){
  result <- "Winner!"
}else{
  result <- "Sorry :("
}
result

[1] "Winner!"

wins <- 0
for(i in 1:1e6){

  ... # insert "switch" strategy here

  if(result=="Winner"){
    wins <- wins + 1
  }
}

wins/1e6

[1] 0.666485

Does this prove our hypothesis? Are you convinced?

What if there were 5 doors? (All the other rules remain the same)

• How often would you win with a “stubborn” strategy?
• How often would you win with a “switch” strategy?

Prove your answers via simulation and send me your code! First one to get it wins… something.

• Take a class on Coursera
• R for Everyone is a great book
• Hang out with me!
• This slideshow with all its links is available at:

rpubs.com/rnorberg/intro-to-R