Summary  
This chapter explains how to detect missing values (`NA`) in tabular data and handle them by replacing with default values or removing rows using data manipulation functions, while considering the impact of each strategy on analysis bias.  

General domain of usage  
data cleaning

Handling missing data is a common challenge in data wrangling with the tidyverse. In R, missing values are represented by the special value `NA`. These `NA` values can arise from incomplete data collection, data entry errors, or merging datasets with non-overlapping entries. If not addressed, `NA` values can disrupt calculations and lead to misleading analysis results. For instance, operations like calculating the mean or sum of a vector containing `NA` will themselves return `NA` unless you explicitly handle the missing values. Recognizing and managing these missing values is essential for ensuring the accuracy and reliability of your data analysis.

options(crayon.enabled = FALSE)
library(tidyverse)

# Creating a tibble with missing values
data <- tibble(
  name = c("Alice", "Bob", "Charlie", "Dana"),
  score = c(95, NA, 88, NA)
)

# Detecting missing values using is.na
missing_scores <- is.na(data$score)

# Replacing missing values with a specific value (e.g., 0) using replace_na
data_filled <- data %>%
  mutate(score = replace_na(score, 0))

print(missing_scores)
print(data_filled)

When handling missing data in your workflow, you should consider both the source of the missingness and the impact of your chosen strategy. Common approaches include:

- Removing rows with missing values;
- Replacing them with a default or imputed value;
- Leaving them as `NA` and using functions that can handle missing values appropriately.

The best practice is to investigate why data is missing and to document the approach you use to address it. In some cases, removing missing values may bias your results, especially if the missingness is not random. Replacing missing values with a constant, such as zero or the mean, may also introduce bias or distort the distribution of your data. Always choose a method that aligns with your analysis goals and the nature of your dataset.

Which of the following statements best describes the implications of different missing data handling strategies in R?

library(testthat)
library(dplyr)

# Load the user's script
source("user_code.R")

# Function for clean output without duplicate test titles or unnamed results
check_logic <- function(test_name, condition, error_msg) {
  test_that(test_name, {
    if (isTRUE(condition)) {
      expect_true(TRUE) # Single success signal
    } else {
      fail(error_msg)
    }
  })
}

# --- TESTS ---

# 1. Missing Temperatures Logical Vector
ok1 <- exists("missing_temps") && 
       is.logical(missing_temps) && 
       length(missing_temps) == 5 && 
       isTRUE(all.equal(missing_temps, c(FALSE, TRUE, FALSE, TRUE, FALSE)))

check_logic(
  "1. missing_temps is a logical vector of correct length and values",
  ok1,
  "Check 'missing_temps'. It should be a logical vector of length 5 indicating where NAs are."
)

# 2. Weather Filled Structure
ok2 <- exists("weather_filled") && 
       inherits(weather_filled, "tbl_df") && 
       isTRUE(all.equal(dim(weather_filled), c(5, 3)))

check_logic(
  "2. weather_filled exists and is a tibble with correct dimensions",
  ok2,
  "Variable 'weather_filled' should be a tibble with 5 rows and 3 columns."
)

# 3. Temperature Replacement Logic
ok3 <- FALSE
if (exists("weather_filled") && exists("weather")) {
  avg_temp <- mean(weather$temperature, na.rm = TRUE)
  expected_temps <- c(25.0, avg_temp, 27.5, avg_temp, 30.0)
  
  ok3 <- isTRUE(all.equal(weather_filled$temperature, expected_temps)) && 
         !any(is.na(weather_filled$temperature))
}

check_logic(
  "3. weather_filled temperature column has NAs replaced with average",
  ok3,
  "NAs in 'weather_filled$temperature' should be replaced by the mean of the original temperatures."
)

# 4. Data Integrity (Original was not modified)
ok4 <- exists("weather") && any(is.na(weather$temperature))

check_logic(
  "4. Original weather tibble remains unmodified",
  ok4,
  "The original 'weather' object should still contain NAs. Do not overwrite it directly."
)

test_main.R

Gain hands-on skills in importing, cleaning, transforming, and preparing data for analysis and visualization in R. This focused course uses the Tidyverse suite to teach practical workflows for real-world data preparation, including handling missing values, data transformation, and building tidy datasets.

Handling Missing Data

解答