Debugging a Tukey Outlier Function in R

 In this exercise, I practiced reproducing and fixing a logical bug in an R function that was supposed to flag rows of a numeric matrix as outliers in every column using the Tukey rule.

First, I ran the original code on a test matrix:

set.seed(123)
test_mat <- matrix(rnorm(50), nrow = 10)
tukey_multiple(test_mat)

The function produced a warning similar to:

'length(x) = 10 > 1' in coercion to 'logical(1)'

The issue came from this line inside the loop:

outliers[, j] <- outliers[, j] && tukey.outlier(x[, j])

The operator && only evaluates the first element of each logical vector and returns a single TRUE or FALSE. That is useful for control flow, but it is incorrect here because the function needs to compare every row element-wise. Since both sides are vectors, the correct operator is &.

I fixed the bug by replacing the line with:

outliers[, j] <- outliers[, j] & tukey.outlier(x[, j])

I also improved the function by adding defensive programming checks to make sure the input is actually a numeric matrix before running the main logic.

Here is the corrected function:

corrected_tukey <- function(x) {
  if (!is.matrix(x)) {
    stop("x must be a matrix.")
  }
  if (!is.numeric(x)) {
    stop("x must be a numeric matrix.")
  }

  outliers <- array(TRUE, dim = dim(x))

  for (j in seq_len(ncol(x))) {
    outliers[, j] <- outliers[, j] & tukey.outlier(x[, j])
  }

  outlier.vec <- logical(nrow(x))
  for (i in seq_len(nrow(x))) {
    outlier.vec[i] <- all(outliers[i, ])
  }

  outlier.vec
}

After rerunning the corrected version on the test matrix, it returned a logical vector of length 10 without errors:

[1] FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE

This debugging process showed the importance of knowing the difference between && and & in R. The bug was not a syntax problem, but a logical one, which makes defensive checks and careful testing especially important.

Comments

Post a Comment

Popular posts from this blog

ABC vs. CBS: Comparing Fictional Presidential Poll Data in R

 # Assignment 3 - Fictional 2016 Presidential Poll Data # Note: This data is made up and does NOT reflect real election results # Candidate names Name <- c("Jeb", "Donald", "Ted", "Marco", "Carly", "Hillary", "Bernie") # Poll results from ABC and CBS ABC_poll <- c(4, 62, 51, 21, 2, 14, 15) CBS_poll <- c(12, 75, 43, 19, 1, 21, 19) # Combine data into a data frame poll_data <- data.frame(Name, ABC_poll, CBS_poll) # Add a column to compare differences between polls poll_data$Difference <- CBS_poll - ABC_poll # View the data print(poll_data) # Optional: Simple bar plot comparison barplot(   t(as.matrix(poll_data[, c("ABC_poll", "CBS_poll")])),   beside = TRUE,   names.arg = poll_data$Name,   col = c("gray70", "gray40"),   legend.text = c("ABC Poll", "CBS Poll"),   main = "Fictional 2016 Presidential Poll Results",   ylab = "Poll Per...
Read more

Assignment #10: Building Your Own R Package

 # Proposal for My R Package: Friedman For this project, I am proposing an R package called **Friedman**. The purpose of this package is to provide beginner-friendly tools for simple data analysis and visualization in R. Many students and new R users struggle with repetitive tasks such as cleaning datasets, calculating summary statistics, and creating clear visualizations. This package will help make those tasks easier by grouping useful functions into one simple package. The main audience for this package is students, beginner analysts, and anyone who wants a more straightforward way to explore data in R. Instead of writing long code repeatedly, users will be able to call functions from Friedman to quickly summarize and visualize their data. This makes the package especially useful for class assignments, small research projects, and practice with R programming. Some of the key functions I plan to implement include: - `clean_data()` – removes missing values or standardizes column n...
Read more

Module # 4 Programming structure assignment

 # ------------------------------------------------------------ # hospital_analysis.R # Hospital patient intake dataset: plots + brief summaries # ------------------------------------------------------------ # ---- 1) Create the dataset ---- Freq <- c(0.6, 0.3, 0.4, 0.4, 0.2, 0.6, 0.3, 0.4, 0.9, 0.2) BP   <- c(103, 87, 32, 42, 59, 109, 78, 205, 135, 176) # first: bad = 1, good = 0, NA = missing First <- c(1, 1, 1, 1, 0, 0, 0, 0, NA, 1) # second: low = 0, high = 1 Second <- c(0, 0, 1, 1, 0, 0, 1, 1, 1, 1) # finaldecision: low = 0, high = 1 FinalDecision <- c(0, 1, 0, 1, 0, 1, 0, 1, 1, 1) hospital_data <- data.frame(Freq, BP, First, Second, FinalDecision) # Print the dataset print("Hospital dataset:") print(hospital_data) # ---- 2) Boxplot: Blood Pressure by Final Decision ---- # Side-by-side comparison of BP for Low vs High final decision boxplot(   BP ~ FinalDecision,   data = hospital_data,   names = c("Low Priority (0)", "High Priority (...
Read more