1 Overview

This extended study guide helps you prepare for the Data Wrangling & Visualization midterm using new practice datasets (distinct from the actual exam). It focuses on: - Cleaning messy data into tidy formats (names, types, text standardization, outliers) - Summaries and grouping (by category and time) - Visualization patterns with ggplot2 - Writing clear insights grounded in evidence from your plots - Responsible use of AI tools for formatting/clarity and debugging, not for full solutions

Assessment weighting: 50% Wrangling, 50% Visualizations & Insights.

2 Packages & Quick Patterns

library(tidyverse)   # dplyr, tidyr, ggplot2
library(janitor)     # clean_names()
library(lubridate)   # dates

Core pipes:

# pattern: read -> clean -> wrangle -> summarize/visualize
## Set your workdirectory to source file location using the session tab on the top menu bar
## save your r file and data files on the same directory
## copy the file path to your desired data file and paste it inside the double quotes for the read_csv command below
## use the commented code chunck below as sample code

# read_csv("path.csv") %>%
#   clean_names() %>%
#   mutate(...) %>%
#   filter(...) %>%
#   group_by(...) %>%
#   summarise(..., .groups = "drop")

3 Practice Datasets (Self-Contained)

Below we create practice-only datasets you can safely use to study.

3.1 Dataset 1: Student Scores

set.seed(123)
n <- 240
students_scores <- tibble(
  student_id   = sprintf("S%03d", 1:n),
  gender       = sample(c("Female","Male"), n, replace = TRUE, prob = c(0.55, 0.45)),
  math_score   = pmax(pmin(round(rnorm(n, 75, 12)), 100), 30),
  reading_score= pmax(pmin(round(rnorm(n, 78, 11)), 100), 30),
  writing_score= pmax(pmin(round(rnorm(n, 76, 13)), 100), 30)
) %>% 
  mutate(across(ends_with("_score"), ~ ifelse(runif(n()) < 0.04, NA, .))) # sprinkle a few NAs

# Optional: save for practice (comment out if not needed)
# readr::write_csv(students_scores, "students_scores.csv")
students_scores %>% head()

Goals to practice with this dataset: - Clean names, compute avg_score - Compare distributions by gender - Identify high performers and potential outliers

3.2 Dataset 2: Bike Sales

set.seed(42)
months <- seq(ymd("2023-01-01"), ymd("2023-12-01"), by = "1 month")
bike_sales <- expand_grid(
  month    = months,
  region   = c("North","South","East","West"),
  bike_type= c("Road","Mountain","Hybrid","Kids")
) %>%
  mutate(
    units_sold  = rpois(n(), lambda = 50) + sample(0:20, n(), replace = TRUE),
    revenue_usd = units_sold * (case_when(
      bike_type == "Road" ~ 850,
      bike_type == "Mountain" ~ 950,
      bike_type == "Hybrid" ~ 650,
      TRUE ~ 300
    )) * runif(n(), 0.85, 1.15)
  )

# Optional: save for practice (comment out if not needed)
# readr::write_csv(bike_sales, "bike_sales.csv")
bike_sales %>% head()

Goals to practice: - Build monthly summaries and category mixes - Create line and stacked area charts - Write short, decision-ready insights

4 Wrangling Fundamentals

4.1 1. Clean Names & Types

scores <- students_scores %>%
  clean_names() %>%
  mutate(
    avg_score = rowMeans(select(., ends_with("_score")), na.rm = TRUE)
  )
scores %>% select(student_id, gender, avg_score) %>% head()

Notes - clean_names(): lower_snake_case - rowMeans() with na.rm = TRUE handles missing scores gracefully

4.2 2. Filtering, Selecting, Arranging

high_achievers <- scores %>%
  filter(avg_score >= 90) %>%
  select(student_id, gender, avg_score) %>%
  arrange(desc(avg_score))
high_achievers %>% head()

4.3 3. Grouped Summaries

by_gender <- scores %>%
  group_by(gender) %>%
  summarise(
    mean_avg = mean(avg_score, na.rm = TRUE),
    sd_avg   = sd(avg_score, na.rm = TRUE),
    n        = n(),
    .groups  = "drop"
  )
by_gender

Try: Add quartiles with quantile(avg_score, probs = c(.25,.5,.75), na.rm = TRUE).

4.4 4. Text Standardization (mini-example)

demo <- tibble(dept = c(" sales ", "Sales", "SALES", " marketing", "Marketing "))
demo %>% mutate(dept_clean = str_to_title(str_trim(dept)))

4.5 5. Date Handling

bike_clean <- bike_sales %>%
  mutate(year = year(month), month_num = month(month))
bike_clean %>% head()

5 Visualization Patterns

Aim for clear titles, labeled axes, readable scales, and captions when needed.

5.1 1. Boxplot (Distribution by Group)

ggplot(scores, aes(x = gender, y = avg_score)) +
  geom_boxplot(outlier.alpha = 0.25) +
  labs(title = "Distribution of Average Scores by Gender",
       x = NULL, y = "Average Score")

Insight prompts - Which group shows larger variability? - Are there extreme outliers? What could cause them?

5.2 2. Bar Chart (Ordered Means)

mean_by_gender <- scores %>%
  group_by(gender) %>%
  summarise(mean_avg = mean(avg_score, na.rm = TRUE), .groups = "drop")

ggplot(mean_by_gender, aes(x = reorder(gender, mean_avg), y = mean_avg)) +
  geom_col() +
  coord_flip() +
  labs(title = "Mean Average Score by Gender", x = NULL, y = "Mean Average Score")

5.3 3. Time Trend (Line)

ggplot(bike_clean, aes(x = month, y = revenue_usd, color = region, group = region)) +
  geom_line(linewidth = 1) +
  labs(title = "Monthly Bike Revenue by Region",
       x = "Month", y = "Revenue (USD)", color = "Region")

5.4 4. Stacked Area (Category Mix Over Time)

by_cat <- bike_clean %>%
  group_by(month, bike_type) %>%
  summarise(revenue = sum(revenue_usd, na.rm = TRUE), .groups = "drop")

ggplot(by_cat, aes(x = month, y = revenue, fill = bike_type)) +
  geom_area() +
  labs(title = "Revenue by Bike Type Over Time",
       x = "Month", y = "Revenue (USD)", fill = "Bike Type")

5.5 5. Small Multiples (Facets)

ggplot(bike_clean, aes(x = month, y = revenue_usd, color = bike_type)) +
  geom_line(linewidth = .9) +
  facet_wrap(~ region, ncol = 2) +
  labs(title = "Revenue by Region and Type",
       x = "Month", y = "Revenue (USD)", color = "Type")

6 Writing Insights (Examples)

Scores (boxplot): “The median average score is similar across genders, but one group shows a wider IQR, indicating more variability.”
Ordered means (bar): “The difference in means is modest; the practical impact depends on grading thresholds.”
Bike trend (line): “Revenue peaks in Q3, suggesting seasonal demand (e.g., summer riding). Consider inventory planning.”
Stacked area: “Mountain and Road dominate revenue; Kids spikes during early summer and holidays.”

Keep insights concise, evidence-based, and tied to charts. Avoid speculation without supporting data.

7 Mini Challenges (Optional)

Scores: Find the top 10% of students by avg_score. Are they concentrated in one gender?
Scores: Create a histogram of avg_score. Experiment with bin widths using bins =.
Bike: Which region has the most volatile revenue? Quantify with standard deviation by region.
Bike: Compare revenue per bike_type using normalized shares (proportion per month).

Sketch of #4:

shares <- by_cat %>%
  group_by(month) %>%
  mutate(share = revenue / sum(revenue, na.rm = TRUE)) %>%
  ungroup()

ggplot(shares, aes(month, share, fill = bike_type)) +
  geom_area() +
  scale_y_continuous(labels = scales::percent) +
  labs(title = "Type Shares Over Time", y = "Share of Monthly Revenue", x = "Month", fill = "Type")

8 Responsible AI Use (Practice)

Allowed use cases: - Debug syntax errors or understand error messages - Improve labels, themes, scales for readability - Brainstorm aesthetic options (not logic)

Not allowed: - Full solutions or writing your entire code

Reflection template (if you used AI): 1. What was the exact question or error? 2. What suggestion did AI offer? 3. What did you keep/modify, and why? 4. How did you verify the result?

9 Suggested Study Routine

Review tidyverse syntax and re-write key commands from memory.
Work through the provided practice datasets to strengthen wrangling skills.
Create several plots (boxplots, line charts, bar charts) and write 2–3 sentence insights.
Run an R Markdown file from start to finish to ensure no errors and reproducibility.
Review the AI use policy and prepare your reflection template.

10 Self-Check Checklist

I can clean names and types; handle NAs and outliers.
I can group and summarise by multiple variables.
I can create boxplots, ordered bars, lines, and stacked area charts.
I can write short, clear insight statements for each plot.
I can knit a document without errors and save figures with ggsave().

11 Appendix: Extra Code Patterns

11.1 Tidy Columns with pivot_longer

scores_long <- students_scores %>%
  clean_names() %>%
  pivot_longer(cols = ends_with("_score"),
               names_to = "subject",
               values_to = "score")
scores_long %>% head()

11.2 Ordering Bars by Value

subject_means <- scores_long %>%
  group_by(subject) %>%
  summarise(mean_score = mean(score, na.rm = TRUE), .groups = "drop")
ggplot(subject_means, aes(x = reorder(subject, mean_score), y = mean_score)) +
  geom_col() +
  coord_flip() +
  labs(title = "Mean Score by Subject", x = "Subject", y = "Mean Score")

End of Study Guide

Study Guide: Data Wrangling & Visualization in R

Abeer Hasan

2025-11-03