facetPlot()

Create Faceted Visualizations of Keyword Trajectories by Zone

The facetPlot() function creates multi-panel plots showing keyword frequency trajectories separated by frequency zones. This faceted approach allows clear comparison of temporal patterns across different vocabulary strata without visual clutter.

🔹 Function Definition

facetPlot(
  data,
  keyword_selection = list(type = "frequency", n = 3, kw.list = NULL),
  r = 4,
  scales = "fixed",
  leg = TRUE,
  themety = "light",
  size_class = NULL,
  x_lab = "years"
)

🎯 Purpose

When visualizing temporal patterns across many keywords, overlapping trajectories can become difficult to interpret. Faceting by frequency zone provides clarity by:

Separating complexity — Each zone displayed in its own panel
Enabling comparisons — Easy to compare patterns across zones
Reducing clutter — Keywords don’t overlap across different frequency ranges
Highlighting representatives — Emphasize specific keywords within each zone
Supporting analysis — Different scales can reveal zone-specific patterns
Creating clarity — Clean, organized visualizations for publications
Facilitating interpretation — Zone-specific dynamics become immediately apparent
Flexible selection — Choose keywords by frequency, randomly, or manually

This function is particularly powerful for corpora with many keywords across diverse frequency ranges.

🧮 Keyword Selection Strategies

The keyword_selection parameter controls which keywords are highlighted in each zone:

1. By Frequency (`type = "frequency"`)

Selects the top N most frequent keywords in each zone.

keyword_selection = list(type = "frequency", n = 3, kw.list = NULL)

Use when: - You want to highlight the most important terms per zone - Exploring core vocabulary within each frequency stratum - Creating representative visualizations

2. Random Selection (`type = "random"`)

Randomly samples N keywords from each zone.

keyword_selection = list(type = "random", n = 3, kw.list = NULL)

Use when: - You want unbiased representation - Exploring general zone behavior - Testing methodology robustness

3. Custom List (`type = "list"`)

Manually specify which keywords to highlight.

keyword_selection = list(type = "list", n = NULL, kw.list = c("algorithm", "data", "network"))

Use when: - You have specific keywords of interest - Creating targeted visualizations - Following up on previous analyses

⚙️ Arguments

Argument	Type	Default	Description
data	List	required	A list object returned by `importData()` or `normalization()`, containing the TDM and corpus metadata.
keyword_selection	List	See below	Controls keyword highlighting: • `type`: `"frequency"`, `"random"`, or `"list"` • `n`: Number of keywords per zone (for frequency/random) • `kw.list`: Vector of specific keywords (for list type)
r	Integer	`4`	Interval for x-axis label thinning. Shows one label every `r` years.
scales	Character	`"fixed"`	Y-axis scale behavior: • `"fixed"`: Same scale across all facets • `"free"`: Each facet has independent scale • `"free_y"`: Free y-axis, fixed x-axis
leg	Logical	`TRUE`	If `TRUE`, displays legend showing zones and highlighted keywords.
themety	Character	`"light"`	Visual theme: • `"light"`: Light background • `"dark"`: Dark background
size_class	Numeric vector	`NULL`	Custom line thickness for each zone. If `NULL`, uses theme defaults.
x_lab	Character	`"years"`	Label for x-axis.

📊 Understanding Scales

Fixed Scales (`scales = "fixed"`)

Zone 4 │ [0-1000]
Zone 3 │ [0-1000]  ← Same y-axis range
Zone 2 │ [0-1000]
Zone 1 │ [0-1000]

Advantages: - Easy to compare absolute frequencies across zones - Immediately see which zones have higher frequencies - Maintains proportional relationships

Disadvantages: - Low-frequency zones may appear flat - Details in smaller zones harder to see

Free Scales (`scales = "free"` or `"free_y"`)

Zone 4 │ [500-1000]
Zone 3 │ [200-500]   ← Each zone optimized
Zone 2 │ [50-150]
Zone 1 │ [0-50]

Advantages: - Each zone’s patterns clearly visible - Reveals details in low-frequency zones - Better for pattern analysis

Disadvantages: - Cannot compare absolute frequencies - May be misleading if not clearly labeled

💡 Usage Examples

Basic Usage - Top Keywords per Zone

library(cccc)

# Import and normalize
corpus <- importData("tdm.csv", "corpus_info.csv")
corpus_norm <- normalization(corpus, normty = "nc")

# Create faceted plot with top 3 keywords per zone
facetPlot(
  corpus_norm,
  keyword_selection = list(type = "frequency", n = 3)
)

Random Selection

# Randomly select 5 keywords per zone
facetPlot(
  corpus_norm,
  keyword_selection = list(type = "random", n = 5)
)

Custom Keyword List

# Highlight specific keywords
facetPlot(
  corpus_norm,
  keyword_selection = list(
    type = "list", 
    kw.list = c("algorithm", "data", "network", "cloud", "machine", "learning")
  )
)

🔍 Interpreting Faceted Plots

What to Look For in Each Panel

High-Frequency Zone Panel

Core vocabulary of the corpus
Typically more stable trajectories
Changes indicate major conceptual shifts
Often shows gradual evolution rather than spikes

Medium-Frequency Zone Panels

Specialized but established terms
May show more dynamic patterns than high-freq zone
Often contains rising/falling terms
Balance between stability and volatility

Low-Frequency Zone Panel

Rare, emerging, or declining terms
More volatile and noisy trajectories
Spikes may indicate temporary importance
Harder to distinguish signal from noise

Cross-Panel Comparisons

Parallel trends across zones: - Indicates corpus-wide phenomena - May reflect historical events or methodological artifacts

Divergent trends: - High-freq stable while low-freq volatile = normal pattern - All zones showing similar patterns = investigate common cause

Zone-specific spikes: - High-freq spike = major event (likely real signal) - Low-freq spike = may be noise or micro-trend

🎯 Choosing Between Fixed and Free Scales

Use Fixed Scales When:

✅ You want to compare absolute frequencies across zones
✅ Showing the magnitude difference between zones is important
✅ Creating visualizations for readers unfamiliar with faceting
✅ Emphasizing that high-freq terms dominate the corpus

Example scenario: Demonstrating that core vocabulary (high-freq) is much more prevalent than specialized terms (low-freq).

Use Free Scales When:

✅ You want to see patterns within each zone clearly
✅ Low-frequency zones would be too flat with fixed scales
✅ Analyzing temporal dynamics rather than absolute frequencies
✅ Looking for similar patterns across different scales

Example scenario: Identifying whether low-frequency emerging terms show similar temporal patterns to established high-frequency terms.

⚠️ Important: Always clearly label when using free scales to avoid misleading readers!

📈 Use Cases

1. Zone Comparison Studies

Compare how different frequency zones evolve over time.

2. Pattern Discovery

Identify zone-specific temporal patterns not visible in combined plots.

3. Publication Figures

Create clear, organized visualizations for papers showing zone structure.

4. Presentation Material

Use free scales to show patterns clearly to audiences.

5. Data Exploration

Systematically examine keywords across frequency strata.

6. Validation

Verify that zone classification captures meaningful differences.

7. Hypothesis Testing

Test whether predicted patterns appear in expected zones.

💡 Tips & Best Practices

Start with frequency selection — See most important terms first
Try both scale types — Compare fixed vs. free to understand your data
Use appropriate n — 2-4 keywords per zone usually optimal
Save large versions — Faceted plots need more space than single plots
Label clearly — Especially important when using free scales
Match theme to context — Light for papers, dark for talks
Consider zone count — More zones = need more vertical space
Test keyword selection — Try random selection to verify representativeness
Document choices — Note which keywords and why in methods
Combine with other plots — Use alongside curveCtuPlot() for comprehensive view

📚 See Also

curvePlot() — Plot all keywords together without faceting
curveCtuPlot() — Zone curves with highlighted examples (alternative approach)
rowMassPlot() — Visualize frequency distribution by zone
normalization() — Normalize before plotting for fair comparison
importData() — Zone classification happens here

💬 Need Help?

For questions about faceted visualizations: - Check the Use Cases page for complete workflow examples - Visit the Projects page to see real applications - Open an issue on GitHub - Contact the team via the About Us page