Readings

This assignment is based on the following readings:

YaRrr: 16, 17

Assignment Goals

Create for loops
Use if statements within for loops
Create custom functions

Examples

# Use a loop to print integers from 1 to 100

for(i in 1:100) {
  
  print(i)
  
}

# Use a loop to add the integers from 1 to 100

sum_total <- 0

for(i in 1:100) {

  sum_total <- sum_total + i  

}


# Create a function showstat that returns a summary statistic from a vector

showstat <- function(x, 
                     what = "mean",
                     na.rm = TRUE) {
  
  if(what == "mean") {output <- mean(x, na.rm = na.rm)}
  if(what == "sd") {output <- sd(x, na.rm = na.rm)}
  if(what == "median") {output <- median(x, na.rm = na.rm)}

  if(what %in% c("mean", "sd", "median") == FALSE) {
    
    stop(paste("Sorry I don't know what", what, "means. Please try 'mean', 'sd', or 'median'")
  }
  
  return(output)

}


# Create a function is.even that tests if an input is even or not

is.even <- function(x,                      # The number to test
                    print.result = FALSE) { # Should the result be printed?
  
  # Test if x is even
even.test <- round(x / 2, 0) == (x / 2)

# Print 'x is even' if even.test is true, otherwise, print 'x is not even!'

if(print.result) {
  
  if(even.test == TRUE) {
    
    message("x is even!")} else {
      
      message("x is NOT even!")}
    
  }

return(even.test)

}

# Try our new function!

is.even(2)
is.even(3)

is.even(2, print.result = TRUE)
is.even(3, print.result = TRUE)

Open your class R project. Open a new script and enter your name, date, and the wpa number at the top. Save the script in the R folder in your project working directory as wpa_9_LastFirst.R, where Last and First are your last and first names.

Loops

Write a loop that prints the integers from 1 to 50 to the console (note: message() works similarly to print()):

for (i in __:__) {
  
  message(__)
  
}

Now repeat the loop, but now print the square of every integer from 1 to 50. That is, it should print 1, 4, 9, 16, …
A participant in one of your studies answered 10 trivia questions. Her score on each question can range anywhere from 0 to 2. The vector 0, 1, 0, 1, 1, 2, 0, 1, 1, 2 represents her score on each of the 10 questions. You want to create a vector that shows the cumulative number of correct questions she answered. The following table shows what I mean

Question	original_score	cumulative_score
1	0	0
2	1	1
3	0	1
4	1	2
5	1	3
6	2	5
7	0	5
8	1	6
9	1	7
10	2	9

Using the following loop, create the vector cumulative_scores that count the cumulative sum of the original scores.

# REPLACE __ WITH THE CORRECT VALUES!
# Vector of original scores on the 10 questions
original_scores <- c(0, 1, 0, 1, 1, 2, 0, 1, 1, 2)

# Create a vector of NAs where the cumulative scores will go
cumulative_scores <- rep(NA, 10)

# Loop over each of the 10 questions
for(i in __:__) {
  
  # Calculate the cumulative sum for the current question
  cumsum_i <- sum(original_scores[1:__])
  
  # Add the value to the ith element in cumulative_scores
  cumulative_scores[__] <- cumsum_i
  
}

# print the result!
cumulative_scores

R has a function called cumsum() that does exactly what your loop does! Try applying cumsum() to the original scores and see if you get the same answer as your loop.

Custom Functions

Write a function called feed.me() that takes a string food as an argument, and prints the sentence “I love to eat food”.

# REPLACE __ WITH THE CORRECT VALUES!

feed.me <- function(___) {
  
  output <- paste0("Yum! I love to eat ", ___)
  
  print(output)
}

Try your function by running feed.me("apples"), it should return Yum! I love to eat apples.
Now adjust your feed.me() function so that if the user specifies “avocados”, then the function returns “NOOOOO, I HATE AVACADOS!”

# REPLACE __ WITH THE CORRECT VALUES!

feed.me <- function(___) {
  
  if(__ != "avacados") {
    
  output <- paste0("Yum! I love to eat ", ___)
  
  }
  
  if( __ == __) {
    
  output <- ""
  
  }
  
  print(output)
}

Write a function called my.mean() that takes a vector x as an argument, and returns the mean of the vector x. Don’t use the mean() function! Use sum() and length()!

# REPLACE __ WITH THE CORRECT VALUES!

my.mean <- function(___) {
  
  result <- sum(___) / length(___)
  
  return(result)
  
}

Try your my.mean() function to calculate the mean weights of chicks in the ChickWeight data frame and compare your results to what you get from using the mean() function.
Write a function called how.many.na() that takes a vector x as an argument, and returns the number of NA values found in the vector)

# REPLACE __ WITH THE CORRECT VALUES!

how.many.na <- function(x) {
  
  output <- sum(is.na(___))
  
  return(___)
}

Test your how.many.na() function to the vector x = c(4, 7, 3, NA, NA, 1)

Checkpoint!

Create your own apa function!

Now it’s time to create your own apa function! Using the following code, write a function called ttest.apa() that takes a numeric vector x, and scaler H0 as arugments, and returns an apa style conclusion from a one-sample test of x against the null hypothesis that the true mean of x is H0.

# REPLACE __ WITH THE CORRECT VALUES!

ttest.apa <- function(x,       # A vector of data
                      mu) {    # The mean under the null hypothesis
  
   # Store the one-sample ttest in object a
  
  a <- t.test(x = ___,    # The vector of data
              mu = ___)   # The mean under the null hypothesis
  
  df <- a$parameter     # Get the degrees of freedom
  test.stat <- ___      # Get the test statistic
  p.value <- ___        # Get the p-value
  
  # If the test is significant
  if(p.value <= ___) {
    
    # Sentence to print for significant result
    
   print(paste0("The test is significant! t(",
                     df, ") = ", test.stat, 
                   ", p = ", p.value, 
                   " (H0 = ", mu, ")"))
  }
  
  # If the test is NOT significant...
    if(p.value > ___) {
     
    # Sentence to print for significant result
      
    print(___)
    
  }
}

Test your ttest.apa() function to see if the following vector of values has a mean that is significantly different from 10. Then, do another test to see if it’s significantly different from 5.

# Is the mean of these values significantly different from 10?
x <- c(6, 8, 12, 6, 8, 3, 5, 3, 7, 0)

Add another argument to ttest.apa() called p.sig that indicates how low a p-value needs to be to be deemed ‘significant’. Then, test your function on the data above, but use the argument p.sig = .001, which means that a p-value must be below 0.001 to be deemed ‘significant’.

Fixing invalid values

One nice application of loops is to sequentially look through columns (or rows) of a dataframe, and make changes. I have a dataset stored at https://raw.githubusercontent.com/ndphillips/IntroductionR_Course/master/assignments/wpa/data/outlier_data.txt. This dataset represents participant’s responses to 100 general knowledge problems. For each question, participants were asked “How confident are you that you know the answer to this question?”. The dataset contains responses from 10 participants (columns) on each of the 100 problems (rows). Load the data into R as a new object called confidence_data.

The data should look like this

p1	p2	p3	p4	p5	p6	p7	p8	p9	p10
75	1000	80	61	67	66	99	85	77	82
81	94	76	70	79	83	80	81	99	75
79	75	89	70	94	77	67	94	83	81
89	88	85	73	70	61	88	89	74	68
81	65	90	84	70	82	88	65	89	88
83	76	70	90	91	76	65	100	82	83

Because the responses are probabilities, none of the participants should have responses smaller than 0 or larger than 100 on any problem. Using the following loop, count the number of times where each participant gave an invalid value:

# REPLACE __ WITH THE CORRECT VALUE(S)!

# Loop over columns
for(i in 1:ncol(__)) {
  
  # Get data from column i
  data.i <- confidence_data[, i]
  
  # Count number of responses below 0 or above 100
  count_invalid <- sum(data.i < __ | data.i > __)
  
  # Print message
  message(paste0("In column ", __, "I found ", __, "values that were either below 0 or over 100"))

}

Now, use the following loop to not only count the number of invalid values, but also to replace the values with NAs

# REPLACE __ WITH THE CORRECT VALUE(S)!

# Loop over columns
for(i in 1:ncol(__)) {
  
  # Get data from column i
  data.i <- confidence_data[,i]
  
  # Count number of responses below 0 or above 100
  count_invalid <- sum(data.i < __ | data.i > __)
  
  # Print message
  message(paste0("In column ", __, "I found ", __, "values that were either below 0 or over 100"))

  # Replace values below 0 or above 100 with NA
  data.i[data.i < __ | data.i > __] <- NA

  # Assign data.i back to the ith column of the data
  __[, i] <- data.i
  
}

Test to see if your code worked! Try running the same loop again over the dataframe and see what you get!

Use a loop to read and write data files

If you have many separate files you want to read into R, you can easily do this using a loop. But before we get to that, we’ll create 100 text files on your computer that, later on, you will read back into R. Of cousre, we will create these 100 text files using a loop! The following loop will create write 100 text files to your working directory. Run it and see what happens!

# JUST RUN! This will create 100 text files in your working directory

for (i in 1:100) {
  
  # Create a random dataset from a hypothetical subject
  data_temp <- data.frame(subject = rep(i, 10),
                          trial = 1:10,
                          response = sample(1:7, size = 10, replace = TRUE))
  
  # Write the data to a text file
  write.table(data_temp, 
              file = paste0("subject_", i, ".txt"), 
              sep = "\t")
  
}

The following code will load each data file into your R session as 100 new objects called subject_1, subject_2, … subject_100. Run the code, then try looking at some of the objects such as subject_40 and subject_64.

# JUST RUN!

for (i in 1:100) {
  
  # Load the data from participant i to a temporary object x
  data_temp <- read.table(file = paste0("subject_", i, ".txt"), 
                          header = TRUE, 
                          sep = "\t")
  
  # Create a new object subject_i containing the data!
  assign(x = paste0("subject_", i), 
         value = data_temp)

}

The code above wrote all of the data to separate objects. But wouldn’t it be better if, instead, the code created one large dataframe containing the data from all participants? The following code chunk will do just that by using the rbind() function. Look at the following code, replace __ with the correct value(s), and then run it to create a new dataframe object called all_data which contains the data from all participants!

# REPLACE __ WITH THE CORRECT VALUE(S)!

# Set up all_data object
all_data <- NULL

for (i in 1:100) {
  
  # Load the data from participant i to a temporary object x
  data_temp <- read.table(file = paste0("subject_", i, ".txt"), 
                          header = TRUE, 
                          sep = "\t")
  
  # Add temprorary data to to all_data!
  all_data <- rbind(__, all_data)

}

Adjust your code above so, for each run of the loop, it prints the message ‘Reading subject X’ (where X is the subject number).

Conduct a simulation!

What is the probability of getting a significant p-value if the null hypothesis is true? Test this by conducting the following simulation:

Create a vector called p.values with 1,000 NA values.
Draw a sample of size 10 from a normal distribution with mean = 0 and standard deviation = 1.
Do a one-sample t-test testing if the mean of the distribution is different from 0. Save the p-value from this test in the ith position of p.values.
Repeat these steps with a loop to fill p.values with 1,000 p-values.

# REPLACE ALL VALUES OF __ WITH THE CORRECT VALUES!

# Create a vector of 1000 NA values
p.values <- rep(NA, ___)

# Loop over all 1000 values
for(i in ___) {

 # Generate a random vector of data
x <- rnorm(n = ___, mean = ___, sd = ___)

# Calculate the te.test
result <- t.test(___)$___

# Store p-value from test in p.values
p.values[___] <- ___

}

What percentage of your 1,000 p values were less than 0.05? In other words, if the null hypothesis is true, what is the probability of getting a p-value less than 0.05?

More data cleaning

The following dataframe survey contains results from a survey of 5 participants. Each participant was asked 5 questions on a 1-10 likert scale. As you can see, many of the responses are not valid integers from 1-10. Using a loop, create a new dataframe called survey.corrected that converts all invalid values to NA:

survey <- data.frame("p" = c(1, 2, 3, 4, 5),
                     "q1" = c(5, 3, 6, 3, 5),
                     "q2" = c(-1, 4, 3, 6, 11),
                     "q3" = c(6, 22, 4, 6, -5),
                     "q4" = c(6, 3, 4, -2, 4),
                     "q5" = c(1, 1, 900, 1, 2))

# REPLACE ALL VALUES OF __ WITH THE CORRECT VALUES!

survey_corrected <- survey   # Copy original survey

for(column.i in ____ ) {  # Loop over columns
 
  x <- ____   # Copy original column vector
  x[(x %in% ___) == FALSE] <- ___  # Replace any bad values
  
  survey_corrected[,___] <- ___ # Assign x back to survey.correced
  
}

Check the values of survey_corrected to make sure it worked!

Submit!

Save and email your wpa_9_LastFirst.R file to me at nathaniel.phillips@unibas.ch.
Go to https://goo.gl/forms/b9dcRH6Ud3pDagOr1 to confirm your assignment submission.

WPA #9 – Loops and Functions

Readings

Assignment Goals

Examples

Loops

Custom Functions

Checkpoint!

Create your own apa function!

Fixing invalid values

Use a loop to read and write data files

Conduct a simulation!

More data cleaning

Submit!