In mattblackwell/qsslearnr: Learnr tutorials based on Quantitative Social Science

library(gradethis)
library(learnr)
library(qsslearnr)
library(tidyverse)
library(stringr)
library(broom)
tutorial_options(exercise.checker = gradethis::grade_learnr)
knitr::opts_chunk$set(echo = FALSE)
tut_reptitle <- "QSS Tidyverse Tutorial 5: Output Report"
set.seed(12345)
n <- 10
x <- sample(100)
y <- x + rnorm(100, mean = 0, sd = 0.5)

Conceptual questions

Loops

quiz(caption = "",
     question("True or False: Loops can simplify our programming code when we want to repeatedly execute code chunks over and over again.",
              answer("True", correct = TRUE),
              answer("False")),
     question("To iterate means to: ",
              answer("perform repeatedly", correct = TRUE),
              answer("perform once"),
              answer("skip over"))
     )

Suppose we assign n <- 10 how many times will the loop for (i in 1:n) iterate?

question_text("Number of iterations:",
              answer("10", correct = TRUE),
              allow_retry = TRUE)

if statement

question("In the conditional expression `if (Y) {expression1}`, `expression1` will be executed if `Y` is:",
         answer("`TRUE`", correct = TRUE),
         answer("`FALSE`"))

question("A conditional statement that is nested within a loop is executed _____ the loop.",
         answer("inside", correct = TRUE),
         answer("outside"))

Coding loops

Loop practice

It is often helpful to create an empty vector to store the computational output from a loop. You can think of this as a bookshelf with spots for the result of each iteration of the loop.

Create a vector called output that is filled with NA values and has length n. We've already loaded n into your environment with a value of 10.

output <-  

grade_result(
  pass_if(~ identical(output, rep(NA, n)), "Good job!")
)

Printing to the screen

In R, it's often very useful to print something to the screen and this is especially helpful when using loops. This can help us diagnose problems in the loop.

Use the print() and str_c functions to put together a phrase and the object n. The final output should read The object n has value 10.

print(str_c("The object n has value ", n))

grade_code()

unique

We've added the following vector fruit to the code. Use the unique function to determine the different types of fruit in the vector. This function will take a vector and return a new vector with the unique values of the original vector.

fruit <- c("apple", "orange", "banana", "orange", "apple", "apple")

fruit <- c("apple", "orange", "banana", "orange", "apple", "apple")

unique(fruit)

grade_code()

FizzBuzz

Here is a variant of the classic "FizzBuzz" coding question. Write a for loop that executes the code

 print(str_c("The following number i is divisible by 3, ", i))

for numbers from 1 to 50 (inclusive) only for the numbers divisible by 3. Include brackets for the conditional. Remember that you can find if a number A is divisible by another number B by using A %% B == 0.

"\n" at the end of print(str_c()) adds a line break to the output.

for(i in 1:50){
  if (i %% 3 == 0) {
    print(str_c("The following number i is divisible by 3, ", i))
  }
}

grade_code()

Adding a condition

Continuing from the previous FizzBuzz code you wrote, use an else if statement to execute the following code if the number is divisible by 5 (but not by 3):

print(str_c("The following number i is divisible by 5, ", i))

for(i in 1:50){
  if (i %% 3 == 0) {
    print(str_c("The following number i is divisible by 3, ", i))
  }
}

for(i in 1:50){
  if (i %% 3 == 0) {
    print(str_c("The following number i is divisible by 3, ", i))
  } else if (i %% 5 == 0) {
    print(str_c("The following number i is divisible by 5, ", i))
  }
}

grade_code()

Linear Regression

Conceptual questions

quiz(
  caption = "",
  question(
    "We use the regression line to predict the value of the outcome variable y hat, otherwise known as:",
    answer("predicted value"),
    answer("fitted value"),
    answer("both of these", correct = TRUE)),
  question(
    "The differences between the observed outcome and its predicted value is called a:",
    answer("residual"),
    answer("prediction error"),
    answer("either of these", correct = TRUE)),
  question(
    "An R^2 close to 1 suggests the model fits:",
    answer("well", correct = TRUE),
    answer("not well")),
  question(
    "With a single binary independent variable, what is the interpretation of the slope coefficient in a linear regression model?",
    answer("Average outcome in the `X = 1` group"),
    answer("Average outcome in the `X = 0` group"),
    answer("Differences in means between the `X=1` and `X=0` groups", correct = TRUE))
)

Interpreting intercepts

In the following box, type answer to the following question: The intercept in the model $Y_i = \alpha + \beta X_i + \epsilon_i$ represents the average value of $Y_i$ when $X_i$ is equal to what?

question_text(
  "The intercept is the average of Y when X is equal to:",
  answer("0", correct = TRUE)
)

Running regressions in R

We have included two vectors x and y. Regress y on x and assign it to an object called fit and print that object to see what the output looks like.

fit <- lm(y ~ x)
fit

grade_result(
  pass_if(~ identical(coef(.result), coef(lm(y~x))))
)

Accessing output from lm

Now, use the coef() function to access coefficients in the fit object.

fit <- lm(y ~ x)

grade_result(
  pass_if(~ identical(.result, coef(fit)), "Good job!")
)

Now you can also access the fitted values using either the fitted() or predict() function. Print the fitted values here:

fit <- lm(y ~ x)

grade_result(
  pass_if(~ identical(.result, fitted(fit)) | identical(.result, predict(fit)),
          "Good job!")
)

Model fit

Use the summary() function to access the value of $R^2$, which is named r.squared inside the summary() output. Remember that you have to access a particular name from that summary using the $.

fit <- lm(y ~ x)

grade_result(
  pass_if(~ identical(.result, summary(fit)$r.squared), "Good job!")
)

Converting the output into tidy data

When working on the output from lm() in tidyverse, you sometimes need to convert the output into tidy data. The broom package from tidymodels provides some useful functions for summarizing the output in the form of tidy data.

Use glance() function to create one-row dataframe summary of the fit object.

fit <- lm(y ~ x)

grade_result(
  pass_if(~ identical(.result, glance(fit)), "Marvelous!")
)

Next, use tidy() function to create a data frame summary in which each row shows a coefficient.

fit <- lm(y ~ x)

grade_result(
  pass_if(~ identical(.result, tidy(fit)), "You nailed it!")
)

Finally, use augment() function to add fitted values, residuals, and other observation level stats to the original data. Here, be sure to add head() to your answer, otherwise, the entire data set will show up.

fit <- lm(y ~ x)

head( )

grade_result(
  pass_if(~ identical(.result, head(augment(fit))), "Keep up the good work!")
)

Submit

submission_ui

submission_server()

mattblackwell/qsslearnr documentation built on Sept. 17, 2022, 6:25 p.m.