Advanced R Topics

Ben Bellman
August 29, 2018

Preface

  • These are important topics, but…
  • Beginning users won't need them right away
    • You have enough to learn in two days as is!
  • Important for advanced programming beyond simple analysis
  • These are foundational, you will need them eventually

"If" and "Else" statements

  • We can use the if ()statement to run code only if a condition is met
x <- 5
if (x < 10) {x + 1}

[1] 6

x <- 15
if (x < 10) {x + 1}

"If" and "Else" statements

  • If statements can be combined with more if statements, or they can be concluded with the else function
x <- 15
if(x < 10){
  x + 1
} else {
  print("Statement was false")
}

[1] "Statement was false"

"While" loops

  • Conditional statements can be used to control loops
  • If the statement is still true at the end of code block, the computer runs the code again
  • When the statement is evaluated as false, the loop stops
x <- 1
while (x < 8){
  x <- x + 1
  print(x)
}

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

"For" loops

  • You can also loop through elements in a vector using the for() function
library(tidyverse)

for(year in 2010:2015){
  paste("The year is", year, sep = " ") %>% 
    print()
}

[1] "The year is 2010"
[1] "The year is 2011"
[1] "The year is 2012"
[1] "The year is 2013"
[1] "The year is 2014"
[1] "The year is 2015"

"For" loops

  • When beginners learn “for” loops, there's a danger they will use them inefficiently
    • My early mistake was looping over rows of data
library(here)
salaries <- read_csv(here("data","white-house-salaries.csv"))

t1 <- Sys.time()
for(a in 1:nrow(salaries)){
  salaries$salary[a] / 1000
}
t2 <- Sys.time()
loop <- t2 - t1
loop

Time difference of 0.0214889 secs

Iteration and Vectorization

  • Loops work through iteration
    • It literally solves one problem at a time as it loops
  • By contrast, vectorized code uses the computer to process things simultaneously
  • dplyr functions are vectorized processes
  • map() from the purrr package in tidyverse vectorizes functions over lists and vectors
  • Only use a loop when you must!

Loop vs. Vectorization

t3 <- Sys.time()
invisible(salaries$salary / 1000)   # suppress output
t4 <- Sys.time()
vect <- t4 - t3

loop

Time difference of 0.0214889 secs

vect

Time difference of 0.002082109 secs

0.01929402 / 0.003953934  # How many times longer was loop?

[1] 4.879702

Writing functions

  • Writing functions has a learning curve, but is rewarding
    • Save time and space
    • Make code easy to share and run
    • Easier to maintain and improve
  • Define with the function() function
  • Arguments define the objects refernced in function code
  • Good idea to keep functions short and nest them

RMarkdown and RStudio

  • I made all these materials inside RStudio!
  • Lots of ways to generate documents with RMarkdown, which is integrated into RStudio development tools
    • Reports, slides, etc. with embedded code and results
  • You can also create websites, like “book sites”, blog posts, and personal sites for professional exposure
    • This is on my to-do list!
    • (working group plug)

Questions?