Lab 3 Walkthrough: Proportions & Hierarchies

STAT 80B Week 5 - Thursday

Prof. Marcela Alfaro Córdoba

Statistics - UCSC

2026-03-10

Welcome to Lab 3!

Today’s Plan

Two big topics:

Part 1: Review Hierarchical Data Concepts

  • What is hierarchical data?
  • Treemaps, Sunburst charts, Mosaic plots
  • When to use which

Part 2: Hands-On Lab 3

  • Create 4 proportion visualizations
  • Build 1 hierarchical visualization
  • Write critical evaluation
  • Work time with help

Lab 3 Overview

Three main parts:

  1. Proportions (4 points) - Pie charts, bar charts, stacked bars
  2. Hierarchies (4 points) - Treemaps, sunburst charts
  3. Critical Evaluation (2 points) - What works? What doesn’t?

Choose your tool: R, Python, or Tableau

Due: End of class today, submit PDF to Canvas

Part 1: Hierarchical Data Concepts

Review: Simple Proportions

Last class: Parts of a simple whole

  • Pie charts: one level of categories
  • Bar charts: comparing categories at same level
  • Stacked bars: comparing groups at same level

Today: What if categories have subcategories?

Hierarchical Data = Nested Levels

Examples in the real world:

  • Taxonomy: Kingdom → Phylum → Class → Family → Genus → Species
  • Organization: Company → Department → Team → Person
  • Geography: Country → State → City → Neighborhood
  • Products: Store → Department → Category → Item
  • Files: Drive → Folder → Subfolder → File

Our Lab 3 Data: Edible Plants

Hierarchical structures we can explore:

  • Taxonomy → Water: Fabaceae → Low/Medium/High water needs
  • Taxonomy → Cultivation: Rosaceae → Fruit Tree, Brassica, etc.
  • Taxonomy → Sunlight: Solanaceae → Full Sun, Partial Shade, Shade
  • Water → Cultivation: High water → what cultivation types?

The challenge: Show multiple levels simultaneously!

The Problem with Simple Charts

Regular pie/bar charts only show ONE level at a time:

We need TWO separate charts to show both levels!

Solution 1: Treemap

Rectangles nested inside rectangles

How to read:

  • Size = count/proportion
  • Nesting = hierarchy levels
  • Color = main categories
  • Borders separate groups

Bigger rectangles = more species

Key features:

  • Space-efficient
  • Shows part-to-whole
  • 2-3 levels work best
  • Easy size comparisons

Treemap Example: Two Levels

Treemap: Strengths & Weaknesses

✅ Advantages:

  • Space-efficient - fits lots of information
  • Shows hierarchy clearly through nesting
  • Big things are literally big (intuitive)
  • Works well with 2-3 levels
  • Good for static displays

❌ Limitations:

  • Hard to compare similar-sized rectangles
  • Some boxes end up awkward shapes
  • Labels can be tricky in small boxes
  • Gets cluttered beyond 3 levels
  • Aspect ratios vary

Solution 2: Sunburst Chart

Nested pies = concentric rings

How to read:

  • Center = root/total
  • Inner ring = top categories
  • Outer rings = subcategories
  • Angle = proportion
  • Interactive = click to zoom

Best for:

  • Beautiful presentations
  • Interactive dashboards
  • Emphasizing proportions
  • Storytelling

Note: Better as interactive visualization!

Treemap vs. Sunburst

Feature Treemap Sunburst
Shape Rectangles Circles
Space use Very efficient Less efficient
Comparison Easier Harder
Aesthetics Functional Elegant
Best for Dense data Presentations
Interaction Helpful Essential

Both are good for hierarchical data - pick based on audience & medium!

Solution 3: Mosaic Plot

Two categorical variables where width AND height matter

How to read:

  • Width ∝ row category size
  • Height ∝ column category size
  • Area ∝ combination frequency
  • Shows relationships!

Good for asking:

  • “Do categories co-occur?”
  • “Are X and Y associated?”
  • “Does distribution vary?”

Not for:

  • True hierarchies
  • Parent-child relationships

Mosaic vs. Treemap

Mosaic Plot:

  • Exactly 2 variables
  • Shows relationships (are X and Y associated?)
  • Not truly hierarchical (no parent-child)
  • Better for independence tests

Treemap:

  • Multiple levels possible
  • Shows hierarchy (parent-child structure)
  • True nesting
  • Better for composition

Use mosaic when asking: “Do cultivation types vary by water requirement?”

Part 2: Lab 3 Hands-On

The Dataset

TidyTuesday Edible Plants Database

What’s in the data:

  • Taxonomy: Plant families (Fabaceae, Rosaceae, Solanaceae, Brassicaceae, etc.)
  • Cultivation: Cultivation classes (Brassica, Legume, Root Vegetable, Leafy Green, etc.)
  • Water: Water requirements (Low, Medium, High)
  • Sunlight: Sun requirements (Full Sun, Partial Shade, Shade)
  • Common and scientific names

This is REAL data about edible plants worldwide!

Loading the Data

library(tidyverse)
library(treemapify)

# Load from TidyTuesday
tuesdata <- tidytuesdayR::tt_load('2026-02-03')
# OR
tuesdata <- tidytuesdayR::tt_load(2026, week = 5)

edible_plants <- tuesdata$edible_plants

# Explore!
glimpse(edible_plants)
names(edible_plants)
# See what's in each column
edible_plants %>% count(taxonomy, sort = TRUE)
edible_plants %>% count(cultivation, sort = TRUE)
edible_plants %>% count(water, sort = TRUE)
edible_plants %>% count(sunlight, sort = TRUE)

# Check for missing data
summary(edible_plants)

First Step: Explore!

Rows: 200
Columns: 6
$ common_name     <chr> "Plant 1", "Plant 2", "Plant 3", "Plant 4", "Plant 5",…
$ scientific_name <chr> "Genus species 1", "Genus species 2", "Genus species 3…
$ taxonomy        <chr> "Fabaceae", "Rosaceae", "Cucurbitaceae", "Apiaceae", "…
$ cultivation     <chr> "Legume", "Brassica", "Grain", "Fruit Tree", "Vine Cro…
$ sunlight        <fct> Partial Shade, Full Sun, Partial Shade, Full Sun, Full…
$ water           <fct> Medium, High, Low, High, Medium, High, High, Medium, M…

Always explore before visualizing!

Proportion Visualizations

The Challenge: Part 1

Create FOUR different visualizations of the same data

Choose ONE approach:

  • Option A: Plant Family Distribution (taxonomy)
  • Option B: Cultivation Types by Water Requirements

All four must show the SAME information, just differently!

Option A: Taxonomy Distribution

Four required visualizations:

  1. Pie chart - proportions as slices
  2. Bar chart - ordered by count
  3. Bar chart - different ordering (alphabetical)
  4. Donut chart - alternative circular

Focus: Which plant families dominate? How concentrated is diversity?

Option B: Water Requirements

Four required visualizations:

  1. Stacked bar chart - cultivation types by water requirement
  2. 100% stacked bar - proportional composition
  3. Side-by-side bars - direct comparison
  4. Small multiples - separate charts per water category

Focus: Does cultivation type relate to water needs?

Pie Chart Example

Bar Chart: Ordered by Count

Stacked Bar Chart Example

Your Turn: Four Proportion Viz (30 min)

Create your four visualizations:

Option A folks:

  1. Pie chart of top families
  2. Bar chart ordered by count
  3. Bar chart alphabetically
  4. Donut chart

Option B folks:

  1. Stacked bar
  2. 100% stacked bar
  3. Side-by-side bars
  4. Small multiples

Each needs: Title, labels, caption with one insight

Hierarchical Visualizations

The Challenge: Part 2

Create ONE hierarchical visualization

Recommended structures:

  • Option 1: Taxonomy → Water requirements
  • Option 2: Taxonomy → Cultivation classes
  • Option 3: Taxonomy → Sunlight requirements

Visualization types:

  • Treemap (recommended)
  • Sunburst chart
  • Icicle plot
  • Circle packing

Treemap Example: Simple

Treemap Example: Two Levels

R Code for Treemaps

library(treemapify)

# Simple treemap (one level)
ggplot(data, aes(area = count, fill = count, label = category)) +
  geom_treemap() +
  geom_treemap_text(colour = "white", place = "centre") +
  scale_fill_gradient(low = "lightgreen", high = "darkgreen")

# Two-level hierarchy
hierarchy_data <- edible_plants %>%
  count(taxonomy, water) %>%
  filter(n >= 2)

ggplot(hierarchy_data, 
       aes(area = n, fill = taxonomy, 
           label = water, subgroup = taxonomy)) +
  geom_treemap() +
  geom_treemap_text(colour = "white", place = "centre") +
  geom_treemap_subgroup_border(colour = "white", size = 3) +
  geom_treemap_subgroup_text(colour = "white", fontface = "bold")

Your Turn: Hierarchical Viz (15 min)

Create ONE treemap showing:

Option 1: Taxonomy → Water requirement

Option 2: Taxonomy → Cultivation class

Option 3: Taxonomy → Sunlight requirement

Requirements:

  • At least 2 levels of hierarchy
  • Clear labeling (use borders and text layers)
  • Strategic color use
  • Professional appearance
  • 2-3 sentence caption

This is worth 4 points, so invest time here!

Tableau Quick Tutorial

For Tableau Users

Quick steps to create treemap:

  1. Load data: Connect to edible_plants.csv
  2. Create hierarchy: Right-click taxonomy → Create Hierarchy → Add water or cultivation
  3. New worksheet: Change mark type to Square
  4. Build treemap:
    • Drag hierarchy to Detail
    • Drag COUNT to Size
    • Drag taxonomy to Color
  5. Format: Add labels, borders, choose colors

Detailed steps in Lab 3 document!

Critical Evaluation

The Writing Component

Write 1-2 paragraphs (200-300 words) addressing:

Section 1: Comparing Your 4 Proportion Visualizations

  1. Which is most effective? Why?
  2. Which is least effective? Why?
  3. For what audience/purpose would each work?
  4. What design choices did you make?

Section 2: Evaluating Your Hierarchical Visualization

  1. What story does it tell about edible plant diversity?
  2. What works well?
  3. What could be improved?
  4. Would a different type be better?

What Makes a Strong Critique?

Good critiques include:

  • Specific examples: “The pie chart works well because Fabaceae dominates at 20%, making the largest slice immediately visible”
  • Trade-offs: “While the alphabetical bar chart makes families easy to find, it obscures which families are most common”
  • Audience thinking: “For botanists, alphabetical might be preferred, but for general audiences, ranked bars communicate better”
  • Design rationale: “I used a green gradient for the treemap to match the plant theme and make larger families darker”

Submission & Working Time

What to Submit

ONE PDF file containing:

  1. Part 1: Four proportion visualizations (1-2 pages)
    • Each with 2-3 sentence caption
  2. Part 2: One hierarchical visualization (1 page)
    • With 2-3 sentence caption
  3. Part 3: Written critique (1 page)
    • Addressing both sections

Total: 3-4 pages

Filename: Lab3_YourLastName.pdf

Grading Breakdown

Component Points Key Criteria
Four proportion visualizations 1.6 All four completed, distinct types
Design quality (proportions) 1.2 Professional, good colors, readable
Labels/captions (proportions) 0.8 Clear titles, axes, insight in caption
Hierarchical viz 4.0 Appropriate type, clear hierarchy, professional
Written critique 2.0 Thoughtful, specific, references concepts

Total: 10 points

Remaining Class Time

Recommended workflow:

  1. ✅ Load and explore the data (5 min)
  2. ✅ Create four proportion visualizations (30 min)
  3. ✅ Create hierarchical visualization (20 min)
  4. ✅ Start written critique (15 min)
  5. ✅ Compile PDF and submit

I’ll circulate to help with:

  • Technical issues
  • Design decisions
  • Interpretation questions

Tips for Success

Data exploration:

  • Check column names first!
  • Count things before visualizing
  • Filter to top 10-15 categories for clarity

Visualization:

  • Less is more
  • Consistent color schemes
  • Clear, large labels
  • Tell a story

Writing:

  • Be specific and honest
  • Use examples from YOUR visualizations
  • Reference class concepts

Questions?

Get started on Lab 3!

I’ll be here to help throughout class.

Remember

There’s no single “correct” visualization. Your critique matters as much as your technical execution!

Thank you!

Happy visualizing! 🌱