In pattiey/STAT545B: Functions for STAT545B

First load all required libraries.

library(tidyverse)
library(datateachr)
library(gapminder)
library(testthat)

Exercise 1: Make a Function

I will make a function that would have been useful in the mini-data analysis. This function takes a tibble and any number of variables as input, then returns the counts for each group in the specified variables, sorted in descending order.

#' Count by group(s)

#' This function returns the counts of specified group(s) within a tibble, 
#' sorted in decreasing order of count
#' 
#' @param data A tibble of interest the contains the data to pull counts from
#' @param ... A list of variables present in the tibble to group-by. When not present, returns the total count in the tibble
#' @return A tibble containing the summarized counts for each group in the variable(s) specified, sorted in decreasing order of count

group_counts <- function(data, ...) {
  stopifnot(is_tibble(data))
  data %>% 
    group_by(...) %>% 
    summarise(count = n()) %>% 
    arrange(desc(count))
}

Exercise 3: Include examples

Show an example to group tree counts by neighbourhood using the vancouver_trees dataset from datateachr

group_counts(vancouver_trees, neighbourhood_name)

Function can also be used with pipe function

vancouver_trees %>% group_counts(neighbourhood_name)

An example grouping by multiple variables in the vancouver_trees dataset

group_counts(vancouver_trees, neighbourhood_name, curb, height_range_id)

The function only works with tibbles, here is an example of it failing when a non-tibble is inputted, and then working with the same dataset as a tibble

group_counts(iris, Species)
group_counts(as_tibble(iris), Species)

Exercise 4: Test the function

Test for the scenarios:

data is a tibble, no variables to group-by (return the total count of the tibble)
data is a tibble, one variable to group-by that is present in the tibble
data is a tibble, multiple variables to group-by that are present in the tibble
data is a tibble, variable to group-by is not present in the tibble
data is a dataframe, function should throw an error
data is not a tibble, function should throw an error

test_that("group_counts function works", {
  expect_equal(dim(group_counts(gapminder)), c(1,1))
  expect_equal(dim(group_counts(gapminder, continent)), c(5,2))
  expect_equal(dim(group_counts(gapminder, continent, year)), c(60,3))
  expect_error(group_counts(gapminder, some_wrong_var))
  expect_error(group_counts(iris, Species))
  expect_error(group_counts(123))
})