Visualizing the Goal

Before you can write effective code, you need to know exactly what you want:

• Goal: Do I want a single value? vector? one observation per person? per year?

• Current State: What do I currently have? matrix, vector? long or wide format?

• Translate: How can I take what I have and turn it into my goal?

  • Sketch out the steps!
  • Break it down into little operations

As we become more advanced coders, this concept is key!!

Remember: When you're stuck, try searching your problem on Google!!

Why Functions?

R (as well as mathematics in general) is full of functions!

We use functions to:

• Compute summary statistics (mean(), sd(), min())
• Fit models to data (lm(Fertility~Agriculture,data=swiss))
• Load data (read_csv())
• Create ggplots (ggplot())
• And so much more!!

Examples of Existing Functions

• mean():
  • Input: a vector
  • Output: a single number

• dplyr::filter():
  • Input: a data frame, logical conditions
  • Output: a data frame with rows removed using those conditions

• readr::read_csv():
  • Input: a file path, optionally variable names or types
  • Output: a data frame containing info read in from file

Each function requires inputs, and returns outputs

Why Write Your Own Functions?

Functions encapsulate actions you might perform often, such as:

• Given a vector, compute some special summary stats
• Given a vector and definition of "invalid" values, replace with NA
• Defining a new logical operator

Advanced function applications (not covered in this class):

• Parallel processing
• Generating other functions
• Making custom packages containing your functions

Anatomy of a Function

NAME <- function(ARGUMENT1, ARGUMENT2=DEFAULT){
  BODY
  return(OUTPUT)
}

• Name: What you call the function so you can use it later

• Arguments (aka inputs, parameters): things the user passes to the function that affect how it works

  • e.g. ARGUMENT1, ARGUMENT2
  • ARGUMENT2=DEFAULT is example of setting a default value
  • In this example, ARGUMENT1, ARGUMENT2 values won't exist outside of the function

• Body: The actual operations inside the function.

• Output: The object inside return(). Could be anything (or nothing!)
  • If unspecified, will be the last thing calculated

Example 1: Doubling A Number

double_x <- function(x){
  double_x <- x * 2
  return(double_x)
}

Let's run it!

double_x(5)

## [1] 10

double_x(NA)

## [1] NA

double_x(1:2)

## [1] 2 4

Example 2: Extract First/Last

first_and_last <- function(x) {
    first <- x[1]
    last  <- x[length(x)]
    return(c("first" = first, "last" = last))
}

Test it out:

first_and_last(c(4, 3, 1, 8))

## first  last 
##     4     8

Example 2: Testing first_and_last

What if I give first_and_last() a vector of length 1?

first_and_last(7)

## first  last 
##     7     7

Of length 0?

first_and_last(numeric(0))

## first 
##    NA

Maybe we want it to be a little smarter.

Example 3: Checking Inputs

Let's make sure we get an error message when the vector is too small:

smarter_first_and_last <- function(x) {
    if(length(x) < 2){
      stop("Input is not long enough!")
    } else{
      first <- x[1]
      last  <- x[length(x)]
      return(c("first" = first, "last" = last))    
    }
}

stop() ceases running the function and prints the text inside as an error message.

Example 3: Testing Smarter Function

smarter_first_and_last(NA)

## Error in smarter_first_and_last(NA): Input is not long enough!

smarter_first_and_last(c(4, 3, 1, 8))

## first  last 
##     4     8

Cracking Open Functions

If you type a function name without any parentheses or arguments, you can see its contents:

smarter_first_and_last

## function(x) {
##     if(length(x) < 2){
##       stop("Input is not long enough!")
##     } else{
##       first <- x[1]
##       last  <- x[length(x)]
##       return(c("first" = first, "last" = last))    
##     }
## }
## <bytecode: 0x7fb785f379b0>

Applying Functions Multiple Times?

Last week, we saw an example where we wanted to take the mean of each column in the swiss data:

for(col_index in 1:ncol(swiss)){
  mean_swiss_col <- mean(swiss[,col_index])
  names_swiss_col <- names(swiss)[col_index]
  print(c(names_swiss_col,round(mean_swiss_col,3)))
}

## [1] "Fertility" "70.143"   
## [1] "Agriculture" "50.66"      
## [1] "Examination" "16.489"     
## [1] "Education" "10.979"   
## [1] "Catholic" "41.144"  
## [1] "Infant.Mortality" "19.943"

Isn't this kind of complex?!

apply(), don't loop!

Writing loops can be challenging and prone to bugs!!

The apply() can solve this issue:

• apply a function to values in each row or column of a matrix
• Doesn't require preallocation
• Can take built-in functions or user-created functions.

Structure of apply()

apply() takes 3 arguments:

1. Data (a matrix or data frame)
2. Margin (1 applies function to each row, 2 applies to each column)
3. Function

apply(DATA, MARGIN, FUNCTION)

For example,

apply(swiss, 2, mean)

##        Fertility      Agriculture      Examination        Education 
##         70.14255         50.65957         16.48936         10.97872 
##         Catholic Infant.Mortality 
##         41.14383         19.94255

Example 1

row_max <- apply(swiss,1,max) #maximum in each row
head(row_max,20)

##   Courtelary     Delemont Franches-Mnt      Moutier   Neuveville 
##        80.20        84.84        93.40        85.80        76.90 
##   Porrentruy        Broye        Glane      Gruyere       Sarine 
##        90.57        92.85        97.16        97.67        91.38 
##      Veveyse        Aigle      Aubonne     Avenches     Cossonay 
##        98.61        64.10        67.50        68.90        69.30 
##    Echallens     Grandson     Lausanne    La Vallee       Lavaux 
##        72.60        71.70        55.70        54.30        73.00

Example 2

apply(swiss,2,summary) # summary of each column

##         Fertility Agriculture Examination Education  Catholic
## Min.     35.00000     1.20000     3.00000   1.00000   2.15000
## 1st Qu.  64.70000    35.90000    12.00000   6.00000   5.19500
## Median   70.40000    54.10000    16.00000   8.00000  15.14000
## Mean     70.14255    50.65957    16.48936  10.97872  41.14383
## 3rd Qu.  78.45000    67.65000    22.00000  12.00000  93.12500
## Max.     92.50000    89.70000    37.00000  53.00000 100.00000
##         Infant.Mortality
## Min.            10.80000
## 1st Qu.         18.15000
## Median          20.00000
## Mean            19.94255
## 3rd Qu.         21.70000
## Max.            26.60000

*Note: Matrix output!

Example 3: User-Created Function

scores <- matrix(1:21,nrow=3)
print(scores)

##      [,1] [,2] [,3] [,4] [,5] [,6] [,7]
## [1,]    1    4    7   10   13   16   19
## [2,]    2    5    8   11   14   17   20
## [3,]    3    6    9   12   15   18   21

my_function <- function(x){ mean(x+10,na.rm=T) }
apply(scores,1,my_function)

## [1] 20 21 22

Activity: My Solution

1. Sort and xtract elements 2 through 6:

   • Answer: Given vector x, use sort(x)[2:6]

2. Function

divers_score <- function(x){
  if(length(x) != 7){
    stop("x is not of length 7!")
  } else{
    x_nofirst_nolast <- sort(x)[2:6]
    return(mean(x_nofirst_nolast))
  }
}

1. Calculate the diver's score given x <- c(2,1:5,3)

divers_score(x = c(2,1:5,3) )

## [1] 2.8

My Answers

1. Preallocate a matrix of NAs with 3 rows and 8 columns, called double_matrix. Manually specify the first column equal to the values 1, 2, and 3. Using a nested loop, fill in the matrix, row by row, such that each value is double that to its left.

double_matrix <- matrix(NA,nrow=3,ncol=8)
double_matrix[,1] <- 1:3
for(row in 1:3){
  for(col in 2:8){
    double_matrix[row,col] <- double_matrix[row,col-1]*2
  }
}
double_matrix

##      [,1] [,2] [,3] [,4] [,5] [,6] [,7] [,8]
## [1,]    1    2    4    8   16   32   64  128
## [2,]    2    4    8   16   32   64  128  256
## [3,]    3    6   12   24   48   96  192  384

My Answers

2. Write an apply() function to take the median value of each column in the cars dataset

median_cars <- apply(cars,2,median)
median_cars

## speed  dist 
##    15    36

My Answers

3. Make a ggplot

library(ggplot2)
ggplot(cars,aes(speed,dist))+geom_point()+
  geom_vline(xintercept = median_cars[1])+
  geom_hline(yintercept = median_cars[2])