Chart Catalogue on Visualizing.JP

Gridded Cartogram

Tue, 31 Mar 2026 00:00:00 +0900

A gridded cartogram represents each region as an equally sized grid cell, such as a square or hexagon, while roughly preserving geographic arrangement. It is also known as a mosaic cartogram, tile grid map, or grid map. Because every region receives the same visual area, small regions are not visually suppressed by large geographic areas.

Historical Background

The gridded cartogram is not tied to a single inventor or paper, but it became especially visible in data journalism and web visualization during the 2010s.

In the United States, tile grid maps representing the 50 states as equal squares became common in election coverage by outlets such as FiveThirtyEight, NPR, and Bloomberg. They helped address a familiar problem: large states such as Alaska, Montana, and Texas dominate ordinary maps, while smaller but often politically important states on the East Coast can nearly disappear.

Similar designs have been used for EU member states, and in Japan for the 47 prefectures. Recent research has also explored algorithms that automatically optimize grid layouts while preserving geographic relationships as much as possible.

Data Structure

The required data is simple compared with many other cartogram types.

Data	Description	Example
Region identifier	Code or name for each region	State code, prefecture code
Grid coordinate	Row and column position	`(row, col)`
Statistical value	Value encoded by color	Population density, turnout, infection rate
Category	Optional value for color grouping	Party, region group
Label	Text shown inside the cell	CA, NY, Tokyo, Osaka

Boundary polygons are not required. The core input is a layout table that assigns each region to a grid position. This layout is often designed manually, although automatic layout methods also exist.

Purpose

The main purpose is to remove the visual bias caused by geographic area. On an ordinary map or choropleth, large territories dominate perception and small territories are easy to miss. A gridded cartogram solves this by giving each region one equal cell. Data values are then represented primarily through color, much like a choropleth map.

Use Cases

State-level election results in the United States
Prefecture-level statistics in Japan
Policy comparisons across EU member states
Country-level indicators such as vaccination rates or education levels
Small multiples showing regional change over time
Sports league comparisons arranged by home location

Characteristics

Every region has equal visual weight.
Geographic arrangement is approximate, not exact.
Data cannot be encoded by area because all cells have the same size.
The technique is easy to implement on the web with tables, CSS Grid, or SVG.
Designing a readable layout can be time-consuming when there are many regions.

How to Read It

Each cell represents one region. Since the cells are equal in size, do not interpret area as magnitude. Read values through color and use the legend to understand the scale or category.

The layout usually preserves a rough north-south and east-west relationship, but adjacency in the grid does not necessarily mean actual geographic adjacency. Labels are therefore important for identifying each region.

Design Notes

Design the layout so readers can infer approximate location.
Add clear labels because shape no longer identifies the region.
Use accessible color schemes, such as ColorBrewer palettes.
Choose squares or hexagons according to the number of regions and layout constraints.
Avoid overloading cells with too many small marks.
Explain the layout logic when the arrangement may not be obvious.

Alternatives

Continuous cartogram: Distorts area in proportion to data while maintaining topology.
Non-contiguous cartogram: Scales each region independently while preserving shape.
Pseudo-continuous cartogram: Replaces regions with circles, squares, or other geometric marks sized by data.
Choropleth map: Uses true geographic shapes and encodes values by color, but area bias remains.

Summary

Gridded cartograms are effective when the goal is to compare regions fairly without letting geographic size dominate perception. They are especially useful in journalism and interactive visualization because they are simple, compact, and easy to scan, provided that the layout and labels are carefully designed.

References

`1`	`- [Cartogram - Wikipedia](https://en.wikipedia.org/wiki/Cartogram)`

Pseudo-Continuous Cartogram

Mon, 30 Mar 2026 00:00:00 +0900

A pseudo-continuous cartogram replaces geographic regions with geometric shapes such as circles, squares, or hexagons, and sizes those shapes in proportion to a data variable. Dorling cartograms and Demers cartograms are representative examples. By discarding the exact shape of each region, the method makes quantitative comparison easier while still roughly preserving geographic position.

Historical Background

The best-known pseudo-continuous cartogram is the Dorling cartogram, proposed by British geographer Daniel Dorling in 1996. Dorling addressed the limitations of continuous cartograms, which distort shapes, and non-contiguous cartograms, which break adjacency, by taking a different route: replace each region with a simple circle.

In a Dorling cartogram, each region is represented by a circle whose area is proportional to the value. A force-directed layout adjusts positions so circles do not overlap while remaining as close as possible to their original geographic locations.

The Demers cartogram later introduced a square-based variant. Squares can produce a more orderly appearance and make labeling easier. Hexagonal variations have also appeared, expanding the range of possible geometric marks.

Data Structure

Data	Description	Example
Region identifier	Code or name for each region	ISO code, prefecture code
Centroid	Representative position used for initial placement	Latitude and longitude
Statistical value	Numeric value that determines area	Population, GDP, medal count
Category	Optional variable for color	Party, region, group

Boundary polygons are not strictly required. Centroids and values are enough to generate a basic version, although boundary information can improve layout quality.

Purpose

The purpose is to preserve geographic context while making quantitative comparison clearer. Complex regional shapes often make area comparison difficult. By replacing all regions with the same type of mark, such as a circle, the reader can compare size more directly.

This is also useful when boundaries are visually complicated or when the exact outline is less important than relative magnitude.

Use Cases

Olympic medal counts by country
Population or GDP comparisons by country
Election results by state or district
Infectious disease case counts by region
Industrial output by prefecture
Branch counts or sales by market area

Characteristics

Area can encode a quantitative value.
Geographic position is approximate but recognizable.
Original shapes and exact adjacencies are lost.
Overlap avoidance and layout algorithms strongly affect readability.
Circles support smooth comparison; squares can support tighter packing and labeling.

How to Read It

Read the size of each circle, square, or hexagon as the data value. Because humans often misread area, the legend should include reference sizes. Color may encode categories or another quantitative measure.

The position of each mark indicates approximate location, not exact geography. If two marks are near one another, that suggests a geographic relationship, but the final layout may be shifted to avoid overlap.

Design Notes

Scale mark area, not diameter, to the data value.
Include a size legend with meaningful reference values.
Keep labels readable and avoid excessive overlap.
Use restrained color so size remains legible.
Explain that the geography is schematic.

Alternatives

Continuous cartogram: Preserves topology but distorts the entire map.
Non-contiguous cartogram: Preserves region shape while scaling each region independently.
Gridded cartogram: Gives every region equal size and encodes values by color.
Proportional symbol map: Places scaled symbols on a normal map rather than replacing regions.

Summary

Pseudo-continuous cartograms are useful when relative magnitude matters more than exact geographic form. By replacing regions with comparable geometric marks, they make values easier to compare while retaining enough spatial arrangement for readers to recognize the geography.

Non-Continuous Cartogram

Sun, 29 Mar 2026 00:00:00 +0900

A non-continuous cartogram scales each region independently in proportion to a data variable while preserving the original shape of that region. Because each region is usually scaled around its centroid, gaps appear between regions and adjacency is not preserved. The advantage is that individual regions remain recognizable, unlike in many continuous cartograms where shapes are heavily distorted.

Historical Background

Non-continuous cartograms were proposed by Judy Olson in 1976 as an alternative to continuous cartograms. Olson focused on the difficulty readers face when regions are distorted so much that they are no longer recognizable. By scaling each region independently, the method makes calculation easier and the result more intuitive.

This approach made the trade-off explicit: lose topological continuity in exchange for shape fidelity and simpler interpretation.

Data Structure

Data	Role
Region geometry	Original polygon shape to be scaled
Region identifier	Key for joining geometry and data
Statistical value	Determines the scaled area
Centroid or anchor point	Center around which the region is scaled
Optional category	Used for color or grouping

Purpose

The purpose is to compare values through area while keeping the recognizability of each region. It is useful when the exact shape of a country, state, or prefecture matters, but geographic adjacency is less important.

Use Cases

Population or GDP by country
Election results by state or prefecture
Regional production or resource comparisons
Public health burden by administrative area

Characteristics

Region shapes are preserved.
Areas encode values.
Adjacency and shared borders are lost.
Gaps between regions are part of the design.
The method is easier to generate than many continuous cartograms.

Design Notes

Make clear which variable controls area.
Use labels because regions move apart visually.
Keep a reference outline or original map nearby if recognition is difficult.
Avoid using the map for distance or adjacency interpretation.

Summary

Non-continuous cartograms are effective when shape recognition and area comparison matter more than adjacency. They provide a simpler and often more readable alternative to continuous cartograms, but the reader must understand that the geography has been separated into independently scaled pieces.

Continuous Cartogram

Sat, 28 Mar 2026 00:00:00 +0900

A continuous cartogram distorts geographic areas in proportion to a data variable while preserving adjacency, or topology. Regions expand or shrink according to values such as population or GDP, producing a map that looks as if a rubber sheet has been stretched. Neighboring regions remain connected, and borders do not break apart.

Historical Background

Cartograms have a long history in thematic cartography, but continuous cartograms became more practical as computational methods improved. The core problem is difficult: change the area of each region to match a value while keeping the map connected and recognizable.

Different algorithms have been proposed to solve this trade-off. Some simulate physical diffusion or rubber-sheet deformation; others optimize geometry to balance area accuracy and shape preservation. The result is always a compromise between statistical accuracy and geographic recognizability.

Data Structure

Data	Description	Example
Boundary geometry	Polygon shapes for each region	Country or prefecture boundaries
Region identifier	Key linking geometry and data	ISO code, prefecture code
Statistical value	Variable used to determine area	Population, GDP, emissions
Optional attributes	Values used for color or labeling	Region group, rate, category

Unlike gridded or Dorling cartograms, a continuous cartogram requires polygon geometry because the shape of the map itself is transformed.

Purpose

The main purpose is to make a map’s visual area correspond to a meaningful data value rather than land area. This is useful when land area is misleading. A population cartogram, for example, enlarges densely populated regions and shrinks sparsely populated ones, making the visual map better match where people actually live.

Use Cases

Population cartograms
GDP or economic output by country
Election maps weighted by population or electorate
Disease burden by region
Carbon emissions and energy consumption
Trade or migration comparisons

Characteristics

Area can represent quantitative values directly.
Topological relationships are preserved.
Shapes can become distorted and unfamiliar.
Small regions may still be hard to label.
Algorithm choice affects both accuracy and readability.

How to Read It

Read the area of each region as the encoded value. A large-looking region is not necessarily geographically large; it has a large value in the chosen data variable.

Because the map remains connected, neighboring relationships are still meaningful. However, distances, angles, and shapes are no longer geographically accurate. Use the legend and title to confirm what variable controls the distortion.

Design Notes

State clearly which variable controls area.
Add labels or reference outlines when distortion makes regions hard to identify.
Avoid mixing area distortion with a confusing color scale.
Use color for a different but related variable only when the relationship is important.
Consider whether the audience can still recognize the geography after distortion.

Alternatives

Pseudo-continuous cartogram: Uses circles or squares sized by value.
Non-contiguous cartogram: Scales regions independently while preserving shape.
Gridded cartogram: Gives regions equal size and uses color for data.
Choropleth map: Preserves geographic shape but is affected by area bias.

Summary

Continuous cartograms are powerful when the map should show the geography of a phenomenon rather than the geography of land. They preserve adjacency while reshaping area to match data, but their effectiveness depends on whether readers can still understand the distorted geography.

Proportional Flow Map

Thu, 26 Mar 2026 00:00:00 +0900

A proportional flow map represents movement between locations by varying line width continuously in proportion to a data value. Migration, trade volume, traffic, and other quantitative flows can be read through the width of the connecting bands. Charles Joseph Minard’s 1869 map of Napoleon’s Russian campaign is one of the most famous examples.

Historical Background

The history of proportional flow maps is closely tied to Minard, a French civil engineer. From around 1845, Minard created maps showing freight movement, population movement, trade, and military campaigns with bands whose widths represented quantity.

His map of Napoleon’s 1812-1813 campaign shows the shrinking size of the army as a proportional band, while also integrating direction, geography, and temperature. Edward Tufte later praised it as one of the finest statistical graphics ever made.

Data Structure

Data	Role
Origin	Starting location
Destination	Ending location
Flow value	Determines line width
Route geometry	Straight or curved path
Category	Optional color or grouping

Purpose

The purpose is to communicate not only where flows go, but also how large each flow is. Continuous width scaling supports more accurate comparison than a purely categorical line symbol.

Use Cases

Migration between regions
Trade flows between countries
Freight and logistics routes
Commuting patterns
River or energy flows
Historical movement narratives

Design Notes

Scale width carefully so large flows do not overwhelm the map.
Use transparency or bundling when many flows overlap.
Clarify whether width represents absolute value, rate, or volume.
Avoid too many crossing lines.
Use arrows only when direction is not otherwise clear.

Summary

Proportional flow maps are powerful when the magnitude of movement is central to the story. They require careful width scaling and route design, but they can reveal both spatial connection and quantitative weight in a single map.

Graduated Flow Map

Wed, 25 Mar 2026 00:00:00 +0900

A graduated flow map represents flows between locations by assigning line widths to discrete classes rather than scaling them continuously. For example, flows may be shown as thin, medium, and thick lines. This reduces visual complexity while still communicating the approximate magnitude of movement, especially when values span a very wide range.

Historical Background

Flow maps date back to the nineteenth century, especially the work of Charles Joseph Minard. The graduated flow map applies the cartographic idea of classification to flow width, just as choropleth maps classify values into color ranges.

As GIS made it easier to draw hundreds or thousands of flows, graduated widths became a practical way to keep maps readable.

Data Structure

Data	Role
Origin and destination	Define the flow
Flow value	Assigned to a class
Classification rule	Equal interval, quantile, natural breaks, or manual
Line style	Width used for each class
Category	Optional color or line type

Purpose

The purpose is to show relative flow magnitude without making every small numeric difference visible. It is useful when the reader needs to understand broad levels rather than exact values.

Use Cases

Regional migration categories
Transportation volume classes
Trade intensity groups
Commuting flows with broad bands
Network flow maps for reports or print

Design Notes

Explain the class breaks.
Keep the number of width classes small.
Use a legend with representative values.
Consider graduated flow when proportional width would create extreme line differences.
Avoid class breaks that hide important thresholds.

Summary

Graduated flow maps trade numeric precision for readability. They are especially useful for dense or wide-ranging flow data where continuous proportional widths would make the map difficult to read.

Spark Words

Tue, 24 Mar 2026 00:00:00 +0900

Spark words visualize quantitative data directly through the typography of inline words. Font weight, size, color, italic style, and other text attributes encode values associated with individual words. Whereas Edward Tufte’s sparklines place tiny charts inside text, spark words make the text itself carry the data. Richard Brath systematizes this form in Visualizing with Text (2020).

Historical Background

Spark words sit at the intersection of two traditions. The first is Tufte’s sparklines, which embed compact data graphics into the flow of prose. The second is typographic visualization, where the visual properties of text are used as data encodings.

Brath’s work extends the idea by treating words as marks. Instead of placing a separate mini-chart next to a word, the word’s own typographic appearance changes according to data.

Data Structure

Data	Role
Word or phrase	Text shown inline
Numeric value	Encoded by weight, size, color, or style
Category	Optional typographic or color grouping
Context text	Sentence or paragraph containing the word

Purpose

The purpose is to preserve reading flow while adding a quantitative layer. Readers can skim text and immediately notice stronger, larger, darker, or differently colored words.

Use Cases

Emphasizing frequency or importance in documents
Showing sentiment or uncertainty in text
Annotating reports without separate charts
Compact dashboards or editorial graphics

Design Notes

Use only a small number of typographic variables.
Keep text readable before adding data encoding.
Avoid making ordinary prose look randomly styled.
Provide a legend or explanation when the mapping is not obvious.

Summary

Spark words are useful when text is the main medium and data should be embedded without interrupting the reader. They work best with restrained typography and a clear mapping between visual style and data.

Text Skimming

Mon, 23 Mar 2026 00:00:00 +0900

Text skimming is a visualization technique that overlays data onto body text through systematic typographic attributes such as bold, italic, font size, color, and underline. It helps readers scan important information without reading every word. It is also called skim formatting, and is discussed by Richard Brath in Visualizing with Text (2020).

Historical Background

Using visual emphasis to guide reading is as old as manuscript and print culture. Decorated initials, headings, bold text, italic emphasis, and color have long helped readers identify structure and importance.

Text skimming differs from ordinary editorial emphasis because the formatting is mapped systematically to data. The visual treatment is not just an author’s judgment; it can encode frequency, sentiment, uncertainty, topic, or another variable.

Data Structure

Data	Role
Text span	Word, phrase, sentence, or paragraph
Data value	Importance, score, frequency, category
Typographic mapping	Font weight, size, color, underline, style
Legend or rule	Explains the mapping

Purpose

The purpose is to combine reading and analysis. A reader can skim a long document and quickly see where important or unusual information is located.

Use Cases

Highlighting key phrases in long reports
Showing sentiment or topic in transcripts
Reviewing survey free-text responses
Annotating legal or policy documents
Educational reading aids

Design Notes

Preserve readability as the first priority.
Use typographic encodings consistently.
Avoid combining too many attributes at once.
Provide a legend when the encoding is data-driven.
Make sure emphasis does not imply importance unless that is the intended meaning.

Summary

Text skimming turns typography into a data layer. It is powerful for document-heavy analysis, but it requires careful restraint so the text remains readable and the visual emphasis remains meaningful.

Typographic Stacked Bar Chart

Sun, 22 Mar 2026 00:00:00 +0900

A typographic stacked bar chart adds data-bearing typography to the labels inside a stacked bar chart. Font weight, size, case, italic style, color, and other text attributes encode dimensions that position and area alone cannot show. Based on ideas discussed in Richard Brath’s Visualizing with Text, the method treats text as a visual mark, not merely as annotation.

Historical Background

Stacked bar charts grew from the broader history of bar charts after William Playfair’s work in the late eighteenth century. They became a common way to show both totals and composition.

Typographic encoding adds another layer to this familiar structure. By mapping values to text attributes, the labels inside segments can communicate additional variables such as emphasis, confidence, growth, or category.

Data Structure

Data	Role
Bar category	Main grouping
Segment category	Parts within each bar
Segment value	Determines segment length
Label text	Displayed inside or near segment
Typographic variables	Encode additional data

Purpose

The purpose is to add information density while keeping the stacked bar structure. It is useful when labels are already necessary and can be designed to carry more meaning.

Design Notes

Keep labels readable in small segments.
Do not encode too many variables through typography.
Use a clear legend for font-based encodings.
Avoid making the chart look like arbitrary styling.
Consider whether a separate chart would be clearer.

Summary

Typographic stacked bar charts can communicate composition and extra label-level variables in one view. They work best when the number of segments is limited and the typographic mapping is simple.

Typographic Mekko Chart

Sat, 21 Mar 2026 00:00:00 +0900

A typographic Mekko chart extends the Mekko chart, or Marimekko chart, by using typographic attributes inside each cell to encode additional dimensions of data. Font size, weight, color, italic style, and other text properties become part of the visualization. In an ordinary Mekko chart, width and height encode two dimensions; in a typographic Mekko chart, the label itself carries more information. This is one form of typographic visualization discussed by Richard Brath in Visualizing with Text (2020).

Historical Background

Mekko charts have long been used in business analysis to show both part-to-whole structure and category composition. The typographic version comes from a broader tradition of treating text not only as annotation but as visual data.

Richard Brath’s work systematized many ways to use text attributes for visualization. The typographic Mekko chart applies that idea to a matrix-like area chart: a cell’s size communicates one relationship, while the appearance of the label communicates another.

Data Structure

Data	Role
Row or segment category	Determines one dimension of the Mekko layout
Column or group category	Determines another dimension of the layout
Cell value	Determines cell area
Text value	Label displayed inside the cell
Typographic variables	Font size, weight, color, style, or spacing

The design requires both numerical values for the area layout and additional attributes that can be mapped to typography.

Purpose

The purpose is to increase the information density of a Mekko chart without adding separate legends, marks, or panels. It is useful when the text labels are already important and can safely carry more meaning.

Use Cases

Market share by segment with growth encoded in label weight
Product portfolio analysis with category size and profitability
Survey results where cell labels show both group names and sentiment
Editorial or text-heavy dashboards where labels are part of the message

Characteristics

It can encode more variables than a standard Mekko chart.
It relies heavily on legible typography.
It can become visually noisy if too many text attributes are used at once.
It works best when the number of cells is moderate.

How to Read It

First read the cell area as in a normal Mekko chart: width and height describe the share or magnitude. Then read the text style. Larger, bolder, darker, or differently colored labels may represent additional values such as growth, importance, risk, or category.

The legend must explain what each typographic attribute means. Without a clear mapping, readers may interpret style as decoration rather than data.

Design Notes

Limit the number of typographic encodings.
Keep labels large enough to read.
Use font weight and color carefully; they attract attention strongly.
Do not encode critical values only through subtle italic or spacing differences.
Check that small cells do not produce unreadable labels.

Alternatives

Standard Mekko chart
Treemap with labels
Heatmap with text annotations
Small multiples of bar charts

Summary

The typographic Mekko chart is a high-density technique that turns labels into data-bearing marks. It is useful when area, category, and text all matter, but it requires careful restraint so the typography remains readable and interpretable.

Typographic Graph

Thu, 19 Mar 2026 00:00:00 +0900

A typographic graph is a network visualization in which nodes are represented by text labels rather than circles or dots. Font size, weight, color, case, and other typographic attributes can encode node properties such as centrality, category, or importance. Edges connect the text labels to show network structure. Richard Brath presents this as a form of typographic visualization in Visualizing with Text (2020).

Historical Background

Network visualization has roots in graph theory, including Euler’s bridges of Konigsberg problem. Modern network visualization expanded rapidly with the internet, social networks, and tools such as Pajek, Gephi, and D3.js.

Traditional network diagrams usually draw nodes as circles and attach labels as secondary information. A typographic graph makes the label itself the node, bringing identity and data encoding into the same visual element.

Purpose

The purpose is to make node identity immediately visible while still representing network relationships. It is useful when names are more important than anonymous points.

Use Cases

Social networks where names or organizations matter
Concept maps
Citation or author networks
Topic networks
Entity relationship diagrams for reports

Design Notes

Prevent text overlap through layout constraints.
Use typography sparingly; too many styles reduce readability.
Keep edge contrast subtle so labels remain legible.
Consider interaction for dense networks.
Use font size carefully because it strongly affects perceived importance.

Summary

Typographic graphs turn labels into primary data marks. They are useful when readers need to identify entities directly, but they require careful layout to avoid clutter and overlapping text.

Parallel Coordinates with River Labels

Wed, 18 Mar 2026 00:00:00 +0900

Parallel coordinates with river labels place text directly along each polyline in a parallel coordinates plot. Standard parallel coordinates can become difficult to read when many lines overlap. By placing labels on the lines themselves, selected data items become easier to follow. Richard Brath discusses related microtext approaches in Visualizing with Text (2020).

Historical Background

Parallel coordinates were developed by Alfred Inselberg and became a standard technique for exploring multidimensional data. Each variable is shown as a parallel axis, and each data record is drawn as a polyline crossing those axes.

The weakness is overplotting. When many records are drawn, individual lines are hard to identify. Color, brushing, and interaction help, but static charts still need clearer identification. River labels borrow from cartographic labeling by attaching text to the path.

Purpose

The purpose is to make important or selected lines traceable without requiring a separate legend or hover interaction.

Use Cases

Comparing named countries, products, or samples across many variables
Highlighting outliers in multidimensional data
Static reports based on parallel coordinates
Educational examples where a few paths must be followed clearly

Design Notes

Use labels only for selected or important lines.
Place text on relatively smooth line segments.
Avoid dense crossings where labels become unreadable.
Use interaction when many labels are needed.
Keep the overall chart light enough for labels to stand out.

Summary

River labels can make parallel coordinates easier to follow by attaching identity directly to the line. The technique is most useful for highlighted records, not for labeling every line in a dense dataset.

Line Chart with River Labels

Tue, 17 Mar 2026 00:00:00 +0900

A line chart with river labels places labels directly along the paths of lines. As river names on maps follow the direction and curvature of rivers, the text in this chart follows the line itself. This reduces the need to look back and forth between the chart and a separate legend. Richard Brath describes related techniques as “microtext lines” in Visualizing with Text (2020).

Historical Background

Direct labeling has long been recommended as a way to make line charts easier to read. The river-label approach extends that principle by bending or positioning text along the line path, borrowing conventions from cartographic labeling.

The idea is especially useful when many lines appear in the same chart. Instead of relying on colors alone, the chart embeds names into the paths, making identification more immediate.

Data Structure

Data	Role
Series identifier	Name displayed as a label
Time or ordered position	X-axis value
Numeric value	Y-axis value
Line geometry	Path used for label placement
Label placement rules	Position, orientation, spacing, and collision handling

The chart requires not only the underlying line data but also enough geometric processing to place labels without making them collide or become unreadable.

Purpose

The purpose is to reduce legend lookup and improve series identification. It is most effective when the viewer needs to follow individual lines across time or compare several trajectories.

Use Cases

Time-series charts with many named countries, companies, or products
Sports ranking trends
Polling or approval-rating lines
Financial or economic indicators
Educational graphics where each line should be read as a named path

Characteristics

Labels are close to the data they identify.
The design reduces reliance on color.
Curved labels can be visually elegant and map-like.
Placement is technically harder than ordinary end labels.
Dense or highly crossing lines can make labels difficult.

How to Read It

Follow the text label along the line to identify the series. Then read the line’s vertical position and slope as usual. The label’s orientation helps the eye stay attached to the correct path.

When labels overlap or are placed only on part of a line, use color or hover interactions as secondary support.

Design Notes

Place labels on relatively smooth segments.
Avoid highly curved or jagged segments that distort letters.
Preserve sufficient contrast between text and background.
Use collision detection or manual adjustment for crowded charts.
Keep a fallback legend or tooltip when the chart is interactive.

Alternatives

End labels at the right edge of a line chart
Direct labels near each line
Traditional legend with color keys
Small multiples

Summary

River-labeled line charts make series identification immediate by putting names directly on the lines. They are strongest when the design has enough space and the lines are smooth enough for labels to remain legible.

Typographic Scatterplots

Mon, 16 Mar 2026 00:00:00 +0900

A typographic scatterplot replaces ordinary point marks with text labels such as words, abbreviations, or names. Each text element directly identifies the data point, reducing the need for tooltips or legends. Font size, weight, italic style, color, and other attributes can also encode additional data dimensions. Richard Brath discusses this technique in Visualizing with Text (2020).

Historical Background

Scatterplots became a basic statistical chart in the nineteenth century. In ordinary scatterplots, each data point is a dot, and identity is added through labels, legends, or interaction.

Typographic scatterplots make the label itself the mark. This idea also resembles cartographic labeling, where place names are positioned as part of the visual structure rather than as afterthoughts.

Purpose

The purpose is to combine position-based comparison with direct identification. It is useful when each point represents an entity readers need to recognize.

Use Cases

Comparing countries, companies, people, or products
Text analysis where words are positioned by two scores
Brand or topic maps
Educational diagrams where identity matters

Design Notes

Avoid too many labels.
Use collision avoidance or selective labeling.
Keep typography legible at the intended size.
Use font attributes consistently and explain their meaning.
Consider ordinary dots plus labels when the chart is dense.

Summary

Typographic scatterplots are useful when identity and position are both important. They can make a scatterplot more self-explanatory, but only if text density is controlled.

3D Yield Curve

Sun, 15 Mar 2026 00:00:00 +0900

A 3D yield curve visualizes changes in bond yield curves over time as a three-dimensional surface. The X-axis represents maturity, the Y-axis represents time, and the Z-axis represents yield. By connecting yield curves from many dates, the chart shows how the term structure of interest rates has evolved.

Historical Background

Yield curves are a standard tool in finance. A two-dimensional yield curve shows interest rates across maturities at one point in time. Analysts have long compared multiple curves to understand changes in monetary policy, inflation expectations, recession signals, and market stress.

The 3D yield curve adds time as an explicit dimension. Instead of comparing several selected lines, it creates a surface that reveals the history of the curve’s shape: steepening, flattening, inversion, and regime shifts.

Data Structure

Data	Description
Date	Observation date
Maturity	Term to maturity, such as 3 months, 2 years, or 10 years
Yield	Interest rate for that maturity and date
Optional metadata	Country, bond type, source, or policy regime

The data is typically a date-by-maturity matrix. Missing maturities may require interpolation.

Purpose

The purpose is to show how an entire yield curve changes over time. A single curve can show one moment, and a line chart can show one maturity over time, but the 3D surface can show both dimensions together.

Use Cases

Treasury yield curve history
Comparison of policy regimes
Recession and inversion analysis
Educational explanations of term structure
Financial dashboards for interest-rate markets

Characteristics

It reveals the overall shape of the yield curve through time.
It makes inversions and steepening visually salient.
It can suffer from occlusion and perspective distortion.
Exact values are harder to read than in 2D charts.
Interactive rotation or slicing can improve usability.

How to Read It

Look along the maturity axis to understand the shape of the yield curve at one date. Look along the time axis to see how that shape changes. Ridges, valleys, and surface color can reveal periods of high or low rates.

If the short end is higher than the long end, the surface shows an inverted yield curve. If long maturities rise far above short maturities, the curve is steep.

Design Notes

Use color carefully to support height, not replace it.
Provide 2D slices or tooltips for exact reading.
Avoid extreme perspective that hides important areas.
Label maturity and date axes clearly.
Consider small multiples when comparison is more important than surface form.

Alternatives

Multiple 2D yield curves
Heatmap of yield by date and maturity
Line chart of selected maturities
Animated yield curve over time

Summary

The 3D yield curve is useful for seeing the evolution of interest-rate term structure as a shape over time. It is visually powerful, but exact analysis usually benefits from accompanying 2D views or interactive inspection.

Heatmap

Sat, 14 Mar 2026 00:00:00 +0900

A heatmap represents values in a matrix or grid by assigning colors to cells. Each cell, at the intersection of a row and column, contains a value, and color intensity or hue communicates its magnitude. Heatmaps are used for correlation matrices, web click analysis, geographic density, gene expression analysis, and many other fields.

Historical Background

The idea of using color to encode quantity is much older than the term heatmap. Nineteenth-century statistical maps and matrix-style graphics already used color to reveal patterns.

The term “heat map” is often attributed to software designer Cormac Kinney, who used it in 1991 for a financial market visualization system that displayed real-time data as a two-dimensional color matrix.

Data Structure

Data	Role
Row category	Y-axis or vertical grouping
Column category	X-axis or horizontal grouping
Cell value	Determines color
Color scale	Maps values to colors
Optional labels	Show exact values or categories

Purpose

The purpose is to reveal patterns across many values at once. A heatmap helps readers see clusters, outliers, gradients, and block structures faster than a table of numbers.

Use Cases

Correlation matrices
Website click and attention analysis
Calendar activity charts
Gene expression matrices
Geographic density surfaces
Business performance dashboards

Design Notes

Choose a color scale that matches the data type.
Use a diverging scale only when there is a meaningful midpoint.
Avoid rainbow scales for ordered values.
Label axes clearly.
Use clustering or sorting when row and column order affects interpretation.

Summary

Heatmaps are effective for scanning large matrices and finding patterns. Their success depends heavily on color scale choice, ordering, and clear legends.

Mekko Chart

Thu, 12 Mar 2026 00:00:00 +0900

A Mekko chart encodes data in both the width and height of rectangular segments. It is also called a Marimekko chart, after the Finnish textile brand whose geometric patterns it resembles. Width often represents an overall category size, while the vertical divisions represent composition within that category.

In statistics, it is closely related to the mosaic plot, and it is useful for showing the relationship between categorical variables and segment scale at the same time.

Historical Background

The academic origin of Mekko charts is linked to mosaic plots. Hartigan and Kleiner introduced mosaic plots in the early 1980s as a way to visualize cross-tabulated categorical data through rectangular areas.

Business use later popularized the Mekko chart for market structure, portfolio analysis, and strategy presentations.

Data Structure

Data	Role
Main category	Determines bar width
Subcategory	Determines segments within each bar
Value	Determines area
Optional color	Encodes subcategory or grouping

Purpose

The purpose is to show both total size and composition in one chart. It answers questions such as: which category is largest, and what is each category made of?

Use Cases

Market share by segment
Product portfolio composition
Revenue by region and product line
Cross-tabulated survey results
Business strategy presentations

Design Notes

Use clear labels because both width and height matter.
Avoid too many categories or small segments.
Make the area encoding explicit.
Consider a stacked bar chart if equal-width comparison is more important.

Summary

Mekko charts are useful when both total category size and internal composition matter. They are information-dense, but they require careful labeling and restraint to remain readable.

Diverging Bar Chart

Wed, 11 Mar 2026 00:00:00 +0900

A diverging bar chart extends bars in two directions from a central baseline, usually zero. It is used to compare positive and negative values, or deviations from a reference point. It is especially useful for Likert-scale survey results, sentiment analysis, and year-over-year performance.

This article focuses on diverging bars where each bar represents a single value, not diverging stacked bar charts where each bar contains multiple segments.

Historical Background

Diverging bar charts grew from the broader history of bar charts after William Playfair’s work in 1786. As statistical graphics developed, analysts needed ways to compare values on both sides of a reference point, especially gains and losses.

Data Structure

Data	Role
Category	One bar
Value	Positive or negative magnitude
Baseline	Usually zero or a meaningful reference
Color	Often indicates direction

Purpose

The purpose is to make direction and magnitude visible at the same time. A reader can quickly see which items are above or below the reference and by how much.

Use Cases

Profit and loss
Sentiment scores
Survey agreement and disagreement
Population change
Performance against target

Design Notes

Use a clear zero or reference line.
Use color consistently for positive and negative values.
Sort bars when ranking is important.
Avoid truncating the axis around zero.
Label the reference value clearly if it is not zero.

Summary

Diverging bar charts are ideal for data with a meaningful center. They make contrast and deviation easy to read, especially when color and ordering are handled carefully.

100% Stacked Bar Chart

Tue, 10 Mar 2026 00:00:00 +0900

A 100% stacked bar chart normalizes every bar to the same total length and compares the proportion of subgroups within each whole. Because each bar adds up to 100%, the chart focuses on composition rather than absolute amount.

It is useful when the question is about share: sales composition by year, energy source mix by region, or response distribution by group.

Historical Background

The 100% stacked bar chart is a variation of the stacked bar chart. It emerged from the broader use of bar charts and statistical graphics as analysts needed to compare proportions rather than raw totals.

Data Structure

Data	Role
Main category	One normalized bar
Subcategory	Segment inside the bar
Value	Converted to a percentage of the bar total
Color	Identifies subcategories

Purpose

The purpose is to compare composition across groups while removing differences in total size. All bars have the same length, so the reader focuses on the internal proportions.

Use Cases

Market share by year
Survey response distribution by group
Energy mix by country
Budget composition by department
Device or platform share over time

Design Notes

Use this chart only when proportions are the main question.
Do not use it when absolute totals matter.
Keep the number of segments limited.
Put the most important segment at a common baseline when possible.
Use clear labels or tooltips for small segments.

Summary

100% stacked bar charts are strong for comparing composition, but they hide total size. They should be used when share matters more than amount.

Stacked Bar Chart

Mon, 09 Mar 2026 00:00:00 +0900

A stacked bar chart divides each bar into multiple segments, allowing the viewer to see both the total value and the composition of that total. Each segment length represents an individual value, while the full bar length represents the sum.

It is widely used for part-to-whole comparisons where both total amount and internal breakdown matter.

Historical Background

Stacked bar charts developed naturally from the bar chart tradition after William Playfair introduced bar charts in the late eighteenth century. As social statistics and business reporting expanded, charts needed to show both totals and components.

Data Structure

Data	Role
Main category	One bar
Subcategory	Segment within the bar
Value	Segment length
Color	Identifies subcategories

Purpose

The purpose is to compare totals while also showing composition. It answers: how large is each group, and what is it made of?

Use Cases

Revenue by business unit and product
Population by region and age group
Budget by department and expense category
Survey responses by group

Design Notes

Segment comparisons are easiest at the shared baseline.
Avoid too many segments.
Use consistent segment order across bars.
Consider a grouped bar chart when comparing subcategories is more important than totals.
Consider a 100% stacked bar chart when proportions matter more than totals.

Summary

Stacked bar charts combine total comparison and composition in one view. They are useful, but segment-level comparison becomes difficult when many colors and categories are involved.

Grouped Bar Chart

Sun, 08 Mar 2026 00:00:00 +0900

A grouped bar chart places multiple data series side by side within each category. It is also called a clustered bar chart. By arranging subgroups next to one another, it allows comparison both across categories and within each category.

Examples include population by gender across regions, or sales by department across years.

Historical Background

Grouped bar charts evolved from the basic bar chart, which William Playfair introduced in the eighteenth century. As comparative statistics expanded, placing multiple bars within a category became a natural way to compare series directly.

Data Structure

Data	Role
Main category	Group on the axis
Series or subgroup	Bars within each group
Value	Bar length
Color	Identifies series

Purpose

The purpose is to compare several series across shared categories. Unlike stacked bars, grouped bars make individual series values easier to compare because each bar has its own baseline.

Use Cases

Sales by year and department
Population by gender and region
Survey scores by group
Product metrics by market

Design Notes

Limit the number of series per group.
Use clear spacing between groups.
Keep series order consistent.
Consider direct labels when legends become burdensome.
Use stacked bars if the total is more important than individual comparison.

Summary

Grouped bar charts are a reliable choice for comparing multiple series within common categories. They become difficult when too many subgroups are added, so restraint is important.

What Is MapLibre Tile (MLT)?

Wed, 28 Jan 2026 00:00:00 +0900

MapLibre Tile (MLT) is a new vector tile format for the MapLibre ecosystem. It builds on the context established by Mapbox Vector Tile (MVT), but is redesigned with modern large-scale geospatial data and next-generation GPU rendering, including WebGPU, in mind. Official MapLibre materials describe support for MLT sources in MapLibre GL JS and MapLibre Native through encoding: "mlt" in a style JSON source.

This article first reviews what vector tiles are, then summarizes the differences between MLT and MVT and how MLT is used in MapLibre.

What Are Vector Tiles?

Vector tiles divide a map into zoom levels and tile coordinates (z, x, y), but instead of serving each tile as an image, they serve geometric features: points, lines, polygons, and their attributes.

Raster tiles contain images such as PNG or JPEG. They are fast to draw, but their style is hard to change after delivery.
Vector tiles contain roads, buildings, points of interest, boundaries, and attributes. The client applies style rules, so color, line width, labels, and visibility can change dynamically.

MVT has been the representative vector tile specification used by many renderers, including MapLibre.

What to Look For

Lines and labels remain crisp while zooming
Vector tiles can change what features and labels are shown at different zoom levels while preserving crisp rendering.
The same data can have different visual styles
Because data and style are separated, a map can be redesigned by changing style rules rather than regenerating all data.
Rendering cost moves to the client
Download size, decoding, filtering, and polygon tessellation can become bottlenecks. MLT is designed to reduce this preprocessing cost.
Complex attributes are increasingly important
Modern geospatial data often has nested attributes, lists, 3D coordinates, and measured values. MLT is designed with these future requirements in view.

MVT Basics

MVT stores tiled vector data efficiently with Protocol Buffers. Its strength is its ecosystem: tools, renderers, debuggers, and production experience. But the format reflects assumptions from roughly a decade ago. With larger datasets and GPU-centered rendering, more work has to be done through schema design and optimization.

What MLT Tries to Improve

MLT is designed around several ideas:

Column-oriented layout: improves compression and filtering by organizing data by columns rather than only by features.
Separation of storage and in-memory formats: supports efficient transfer while allowing faster runtime processing.
GPU-friendly design: supports moving expensive work such as polygon tessellation earlier in the pipeline.
Richer types: anticipates complex attributes, 3D coordinates, and measured values.

MVT and MLT Compared

Aspect	MVT	MLT
Basic idea	Established vector tile format	Redesigned for next-generation rendering
Storage model	Feature-oriented	More column-oriented
Decoding and processing	Mature ecosystem, but heavy data can be costly	Designed for faster compression, decoding, and filtering
Polygon preparation	Runtime tessellation can be a bottleneck	Supports pre-tessellated storage concepts
Attribute types	Mostly simple types	Designed with complex and future types in mind

MapLibre Support

MapLibre documents MLT as a supported vector source encoding. A style can specify encoding: "mlt" in a vector source.

{
  "version": 8,
  "sources": {
    "basemap": {
      "type": "vector",
      "url": "https://example.com/tiles/style-or-tileset.json",
      "encoding": "mlt"
    }
  },
  "layers": [
    {
      "id": "roads",
      "type": "line",
      "source": "basemap",
      "source-layer": "transportation"
    }
  ]
}

MapLibre GL JS release notes and examples also describe MLT support, making it a practical part of the MapLibre stack rather than only a specification proposal.

Why MLT Now?

The assumptions behind vector tiles are changing. Data volume is growing, browsers can use the GPU more effectively, and rendering stacks increasingly benefit from data structures that are easier to transfer and process in parallel. MLT responds to this shift by treating the format itself as part of the performance solution.

Summary

Vector tiles serve map features instead of images, allowing dynamic styling. MVT made this approach mainstream. MLT is MapLibre’s attempt to modernize the format for larger data, faster decoding, better filtering, and GPU-friendly rendering. Its use through encoding: "mlt" positions it as a concrete evolution of the MapLibre ecosystem.

References

Isoline Map

Wed, 10 Dec 2025 00:00:00 +0900

An isoline map visualizes continuous spatial change by connecting points with the same value. Contour lines, isotherms, and isobars are familiar examples. Unlike choropleth maps, which color administrative areas, isoline maps emphasize the continuity of the phenomenon itself.

Historical Background

The idea of isolines is old. Edmond Halley produced an early scientific isoline map in 1701 showing magnetic declination in the Atlantic. In the eighteenth and nineteenth centuries, contour lines and isotherms became important tools in topography, meteorology, oceanography, and geology.

With GIS, isolines can now be generated from observation points using interpolation methods such as kriging, inverse distance weighting, and splines.

Data Structure

Data	Role
Observation points	Locations with measured values
Values	Temperature, elevation, pressure, travel time, etc.
Interpolation method	Generates a continuous surface
Contour interval	Determines which values are drawn as lines

Purpose

The purpose is to reveal gradients and spatial structure. Isolines show how values change across space, where peaks and valleys occur, and how smoothly a phenomenon varies.

Use Cases

Elevation contours
Weather maps
Sea-level or ocean temperature maps
Air pressure and wind analysis
Travel-time accessibility maps

Design Notes

Choose contour intervals carefully.
Label lines when exact values matter.
Avoid drawing too many lines.
Combine with color only when it improves clarity.
Note the interpolation method when values are estimated.

Summary

Isoline maps are effective for continuous phenomena. They help readers see gradients and patterns, but the result depends strongly on interpolation and contour interval choices.

Isopleth Map

Wed, 10 Dec 2025 00:00:00 +0900

An isopleth map is a thematic map that represents the distribution of a continuous quantity by coloring areas bounded by lines of equal value. Observation points are interpolated into a continuous surface, and zones between equal-value boundaries are filled with color or shading. It is useful for seeing phenomena such as temperature, humidity, air pollution, and population density as spatial fields.

Historical Background

The term “isopleth” is associated with a classification proposed by John K. Wright in 1944. Wright distinguished lines connecting directly measurable point values from lines representing values such as ratios or densities that are derived for areas. This distinction helped formalize contour-based thematic mapping.

From the 1960s onward, computer-based interpolation made it easier to extract isolines and fill the resulting areas. With the spread of GIS in the 1990s, isopleth maps became standard in meteorology, environmental monitoring, and spatial analysis.

Data Structure

Data element	Description	Example
Observation coordinates	Locations of measurement points	Weather stations, air-quality monitors
Observed value	Numeric value at each point	Temperature, PM2.5, population density
Interpolation method	Algorithm for estimating the surface	Kriging, IDW, spline
Classification method	How value ranges are divided	Equal interval, quantile, Jenks
Color scheme	Colors assigned to classes	Sequential gradient, stepped palette

Purpose

The purpose is to communicate the spatial spread and intensity of a continuous variable through color-filled areas. Compared with an isoline map, which uses only lines, an isopleth map gives a stronger visual impression of high and low regions.

Use Cases

Weather maps for temperature, humidity, and wind speed
Air-pollution maps for PM2.5, ozone, or NO2
Urban heat-island analysis
Ocean maps for sea surface temperature and salinity
Noise or radiation exposure mapping

Characteristics

Aspect	Description
Target	Spatially continuous quantities
Input	Discrete observation points plus interpolation
Representation	Colored areas bounded by equal-value lines
Strength	Makes broad distribution patterns easy to see
Weakness	Depends heavily on interpolation and classification
Tools	ArcGIS, QGIS, Python, R

How to Read It

First check the legend to understand the color order and units. Then note how the classes were created, because equal intervals, quantiles, and natural breaks can produce different impressions from the same data.

Remember that the map may show estimated values, not direct observations, especially in areas far from measurement points. If possible, inspect the observation locations as well.

Design Notes

Use a sequential palette for one-directional values and a diverging palette when there is a meaningful midpoint.
State the classification method and number of classes.
Use color-vision-friendly palettes.
Include units, interpolation method, data source, and uncertainty notes.
Keep area boundaries subtle so color remains the main signal.

Alternatives

Method	Feature	Suitable when
Isoline map	Shows equal-value lines only	You need to read gradients or structure precisely
Heat map	Colors pixels continuously	A smoother surface is desired
Choropleth map	Colors administrative areas	The purpose is administrative comparison
Raster map	Uses grid cells	Resolution should be explicit

Summary

Isopleth maps are useful for showing continuous spatial phenomena as colored surfaces. They are common in weather, environmental, and urban analysis, but their credibility depends on appropriate interpolation and classification.

References

1
2

- [Contour line - Wikipedia](https://en.wikipedia.org/wiki/Contour_line)
- [Isarithmic map - Wikipedia](https://en.wikipedia.org/wiki/Isarithmic_map)

Scatterplot Matrix

Wed, 12 Nov 2025 00:00:00 +0900

A scatterplot matrix, or SPLOM, displays pairwise relationships among multiple variables in a grid of scatterplots. Each variable is combined with every other variable across rows and columns, allowing correlations, patterns, clusters, and outliers to be explored at once.

Historical Background

Scatterplot matrices developed in the context of exploratory data analysis, especially the work of John W. Tukey. They became standard in statistical software such as S and later R, and are now available in Python libraries, Tableau, Power BI, and other tools.

Data Structure

The data is a table with observations as rows and numeric variables as columns. A matrix is generated by plotting each variable against every other variable.

Purpose

The purpose is to examine many pairwise relationships quickly. It is especially useful at the beginning of analysis, before deciding which relationships deserve more detailed modeling.

Design Notes

Use it for a moderate number of variables.
Add color for categories when helpful.
Consider histograms or density plots on the diagonal.
Use brushing or interaction for dense datasets.
Avoid overplotting through transparency or sampling.

Summary

Scatterplot matrices are a basic tool for multivariate exploration. They help analysts see relationships and outliers across many variables, but they become difficult to read when the number of variables or observations is too large.

Word Bubble

Wed, 12 Nov 2025 00:00:00 +0900

A word bubble chart represents words from text data as circles. The size and color of each bubble can encode frequency, importance, category, or sentiment. It resembles a word cloud, but avoids overlapping words and can provide a clearer layout and stronger sense of relative structure. It is often used to communicate results from natural language processing and text mining.

Historical Background

Word bubbles emerged as a more structured extension of word clouds. From the 2010s onward, implementations appeared in visualization tools and D3.js examples under names such as word bubble chart or bubble cloud. The method became useful in data journalism, education, marketing, and survey analysis because it combines visual appeal with clearer comparison.

Data Structure

Field	Example	Description
`word`	“data”	Word to display
`frequency`	120	Count or score
`category`	“technology”	Optional topic or group

Bubble size is usually determined by frequency, while color often represents category or sentiment.

Purpose

The purpose is to make important or characteristic words in a text collection visible at a glance. Because relative frequency can be compared visually, the method is useful for presenting quantitative text analysis in an intuitive way.

Use Cases

Sentiment analysis of social media and reviews
Summaries of free-text survey responses
Keyword analysis of speeches and articles
Brand and product language analysis
Vocabulary exploration in education

Characteristics

More structured than a word cloud because overlaps are avoided.
Size and color can encode multiple dimensions.
Interactive tooltips and clicks can show details.
It can balance visual appeal and readability when the word count is controlled.

How to Read It

Bubble size shows word frequency or score.
Bubble color often shows sentiment or category.
Placement may reflect frequency, semantic grouping, or clustering.
Labels identify the words and make comparison possible.

Design Notes

Do not show too many words.
Use appropriate scaling so smaller terms remain readable.
Choose colors for category distinction and accessibility.
Avoid random-looking layouts when semantic or frequency structure matters.

Alternatives

Method	Feature	Suitable when
Word cloud	Quick overall impression	You want a broad textual impression
Heatmap	Precise value comparison	Quantitative differences matter
Network diagram	Shows relationships	Co-occurrence and context matter
Circle packing	Preserves hierarchy	Words are grouped by categories

Summary

Word bubbles are a structured way to visualize important words in text data. Compared with word clouds, they provide better layout control and comparison, making them useful for journalism, education, marketing, and text analytics.

References

Word Cloud

Wed, 12 Nov 2025 00:00:00 +0900

A word cloud visualizes words from text data by changing font size, color, or placement according to frequency or importance. It is often used to give a quick impression of documents, social media posts, survey responses, and other text collections. Larger words usually indicate higher frequency or importance.

Historical Background

Word clouds grew from tag clouds, which became common on blogs and web services in the early 2000s. Tag clouds displayed tags according to frequency. Later, natural language processing and visualization tools extended the idea to words extracted from arbitrary text.

IBM’s Wordle, created by Jonathan Feinberg in 2008, helped popularize the modern word cloud.

Data Structure

Data	Role
Word	Text to display
Frequency or score	Determines size
Category or sentiment	Optional color grouping

Purpose

The purpose is to provide a quick visual summary of prominent terms. It is useful for exploratory or communicative contexts, but it is not ideal for precise comparison.

Design Notes

Remove stop words and irrelevant terms.
Avoid showing too many words.
Use color with meaning, not only decoration.
Do not overclaim precise quantitative findings from layout.
Consider bar charts when exact ranking matters.

Summary

Word clouds are accessible and memorable, but they are approximate. They work best as an entry point into text data rather than as a precise analytical chart.

3D Scatterplot

Sun, 12 Oct 2025 00:00:00 +0900

A 3D scatterplot places points in three-dimensional space using X, Y, and Z axes to represent three numeric variables. It extends the ordinary two-dimensional scatterplot by adding depth. Interactive rotation and zoom can help reveal clusters or structures that are hard to see from a single view.

Historical Background

Scatterplots became standard with the development of statistics. Practical 3D scatterplots became common only after computer graphics and interactive statistical software matured in the late twentieth century. Today they are supported by tools such as Matplotlib, Plotly, R, Tableau, and WebGL-based libraries.

Data Structure

Data	Role
X value	Horizontal position
Y value	Vertical position
Z value	Depth or height
Optional category	Color or symbol

Design Notes

Use interaction when possible.
Be careful with occlusion and perspective distortion.
Provide 2D projections or slices for precise reading.
Avoid using 3D when a 2D scatterplot or matrix is clearer.

Summary

3D scatterplots can reveal multidimensional structure, but they are harder to read precisely than 2D views. They work best in interactive contexts.

Bar Chart

Sun, 12 Oct 2025 00:00:00 +0900

A bar chart compares values across categories using the length of bars. One axis contains categories, and the other contains numeric values such as counts, amounts, or percentages. Bar charts are among the most basic and widely used visualization methods.

Historical Background

The bar chart is usually traced to William Playfair’s The Commercial and Political Atlas (1786). Playfair used bars as a clear way to compare quantities visually, helping establish modern statistical graphics.

Data Structure

Data	Role
Category	One bar
Value	Bar length
Optional group	Color or grouped/stacked structure

Purpose

The purpose is direct comparison. Bar length is easier to compare than area, angle, or color intensity, making bar charts reliable for many audiences.

Design Notes

Start the value axis at zero in most cases.
Sort bars when ranking matters.
Use horizontal bars for long labels.
Avoid unnecessary 3D effects.
Use color sparingly and meaningfully.

Summary

Bar charts are a default choice for comparing category values because they are simple, accurate, and easy to read.

Bubble Chart

Sun, 12 Oct 2025 00:00:00 +0900

A bubble chart is an extension of the scatter plot that can represent relationships among three or more variables in one chart. The X- and Y-axes show two quantitative variables, while the area of each circle, or bubble, represents a third quantitative variable. Color, opacity, or shape can add further information.

Bubble charts are widely used for business analysis, economic comparison, marketing analysis, and other multivariate datasets.

Historical Background

The bubble chart traces its roots to the scatter plot. By the 1960s and 1970s, statistical and graphics systems were using point size to add another quantitative dimension to scatterplots.

The format became widely known through Hans Rosling’s Gapminder project, which used animated bubble charts to show global health and economic data. This popularized the animated bubble chart, or motion chart, as a symbol of data storytelling.

Data Structure

Category	X value	Y value	Size value	Color
Country A	50000	75	1.2	Blue
Country B	30000	65	0.8	Red

X-axis and Y-axis: numeric variables, such as GDP and life expectancy.
Size: a third variable, such as population.
Color: an optional category such as region, industry, or gender.

Purpose

The purpose is to compress a multivariate relationship into a two-dimensional space while preserving a sense of magnitude. A scatter plot shows the relationship between two variables; a bubble chart adds weight or scale.

Use Cases

Economic analysis: GDP, life expectancy, and population by country
Marketing: sales, profit margin, and market share
Education data: test scores, investment, and population share
Sustainability: CO2 emissions, renewable energy share, and population

Gapminder’s world development data is the classic example: GDP on the X-axis, life expectancy on the Y-axis, and population as bubble size.

Characteristics

Shows more information than a scatter plot.
Can compare categories through color or shape.
Overlapping bubbles can reduce readability.
Size scaling must be handled carefully.

How to Read It

Position: Read the relationship between the X and Y variables.
Bubble size: Larger bubbles indicate larger values, but area perception is imprecise.
Color: Indicates group or region.
Animation: If present, shows change over time.

Design Notes

Scale bubble area, not diameter, to the value.
Use transparency or layout techniques to reduce overlap.
Give color a clear meaning and include a legend.
Avoid 3D bubbles and perspective effects.
Add labels selectively to avoid clutter.

Alternatives

Purpose	Alternative
Correlation among many variables	Scatterplot matrix
Weighted distribution	Heatmap
Part-to-whole comparison	Treemap
Time-series comparison	Line chart or sparklines

Summary

Bubble charts are powerful for showing multivariate relationships, especially when size itself is meaningful. They require careful scaling and design because area perception and overlap can easily mislead readers.

References

How to Choose Among Bar Chart Variants

Sun, 12 Oct 2025 00:00:00 +0900

Bar charts are familiar, but choosing the wrong bar chart variant can weaken the message. This article organizes four basic bar-chart forms using the ideas of dimensions and measures.

Dimensions and Measures

A dimension is a categorical field such as region, product, or year. A measure is a numeric value such as sales, population, or percentage. Bar charts compare measures across dimensions.

Four Common Forms

Simple bar chart: compares one measure across one dimension.
Grouped bar chart: compares multiple series within each category.
Stacked bar chart: shows totals and composition.
100% stacked bar chart: compares composition after normalizing totals to 100%.

Choosing the Right One

Use a simple bar chart when ranking or magnitude comparison is the goal. Use grouped bars when comparing subgroups directly. Use stacked bars when totals and parts both matter. Use 100% stacked bars when proportions matter more than absolute totals.

Summary

Bar charts are not one chart but a family of related forms. The right choice depends on whether you want to compare totals, subgroups, composition, or proportions.

Line Chart

Sun, 12 Oct 2025 00:00:00 +0900

A line chart connects points with lines to show numerical change. It is most often used for time series, with time or ordered sequence on the X-axis and quantity on the Y-axis. It makes trends, increases, decreases, cycles, and turning points easy to see.

Historical Background

Line charts date back to the late eighteenth century. William Playfair used line charts in The Commercial and Political Atlas in 1786 to show trade data over time. Since then, line charts have become one of the most universal forms in science, government, journalism, and business.

Data Structure

Data	Role
Time or sequence	X-axis
Numeric value	Y-axis
Series	Optional multiple lines

Purpose

The purpose is to show change over ordered values, especially time. Line charts are strongest when the continuity or trajectory between points matters.

Design Notes

Use line charts for ordered or continuous X-values.
Do not connect unrelated categories.
Label axes and units clearly.
Use direct labels when multiple lines are present.
Avoid too many overlapping series.

Summary

Line charts are the standard chart for showing change over time. Their strength is continuity: they let readers see direction, slope, and pattern at a glance.

Pie Chart

Sun, 12 Oct 2025 00:00:00 +0900

A pie chart represents parts of a whole as slices of a circle. The whole is treated as 100% or 1, and each category’s share is shown by the angle and area of its slice. Pie charts are intuitive for simple composition, but they are not very precise for comparison.

Historical Background

The pie chart was introduced by Scottish political economist William Playfair in 1801 in Statistical Breviary. Florence Nightingale and other reformers later helped popularize circular statistical graphics.

Data Structure

Data	Role
Category	One slice
Value	Slice angle and area
Total	Sum of all slices

Purpose

The purpose is to show a simple part-to-whole relationship. It works best when there are only a few categories and one or two slices are clearly dominant.

Design Notes

Avoid too many slices.
Do not use pie charts for precise ranking.
Label values directly when possible.
Avoid 3D or exploded effects.
Consider a bar chart when comparison matters.

Summary

Pie charts are familiar and useful for simple composition, but they should be used carefully. They communicate “share of whole” well, but they are weak for precise comparison among similar values.

Range Plot

Sun, 12 Oct 2025 00:00:00 +0900

A range plot connects two values for each item with a line, making the range between them visible. It is used when each category has a lower and upper value, a start and end, or two comparable values such as last year and this year. It is closely related to the dumbbell chart and gap chart.

Historical Background

Range plots developed from dot plots. William Cleveland formalized dot plots in the context of statistical graphics, and connecting two points with a line became a natural way to show gaps and intervals.

Purpose

The purpose is to make differences between two values easy to compare across categories. The length of the line shows the size of the gap, while the positions of the endpoints show the actual values.

Use Cases

Minimum and maximum values
Before-and-after comparison
Gender gaps
Year-over-year change
Start and end dates or values

Design Notes

Label endpoints clearly.
Sort categories by gap or endpoint when helpful.
Use color to distinguish the two endpoint types.
Avoid too many categories if labels become crowded.

Summary

Range plots are effective when the story is about the interval between two values. They are more precise than grouped bars for showing gaps and often more compact.

Scatter Plot

Sun, 12 Oct 2025 00:00:00 +0900

A scatter plot is a fundamental statistical chart for showing the relationship between two numeric variables. One variable is assigned to the X-axis and the other to the Y-axis, and each observation is plotted as a point. The distribution of points reveals correlation, trends, clusters, and outliers.

Historical Background

Scatter plots became important in the nineteenth century through the work of statisticians and scientists such as John Herschel and Francis Galton. Galton’s studies of height, correlation, and regression helped establish the analytical value of point-based comparison.

Data Structure

Data	Role
X value	Horizontal position
Y value	Vertical position
Observation	One plotted point
Optional category	Color or shape
Optional size	Additional quantitative variable

Purpose

The purpose is to understand how two variables move together. A scatter plot can show positive correlation, negative correlation, nonlinear structure, clusters, or unusual observations.

Design Notes

Start axes at meaningful ranges rather than automatically forcing zero.
Use transparency when points overlap.
Add trend lines only when they support the analysis.
Use color for categories, not decoration.
Consider a scatterplot matrix for many variables.

Summary

Scatter plots are one of the most important tools in exploratory data analysis. They are simple, flexible, and effective for revealing relationships between numeric variables.

The Crocodile-Mouth Chart: A Japanese Metaphor for Fiscal Imbalance

Sun, 12 Oct 2025 00:00:00 +0900

The “crocodile-mouth chart” is a distinctive line-chart metaphor used in Japan to explain the country’s fiscal situation. It shows government expenditure and tax revenue as two lines. As the gap between them widens, the shape resembles a crocodile opening its mouth.

This expression is specific to Japanese media and policy discourse. It is not a standard chart type commonly known in English as a “crocodile chart.”

How to Read It

The upper line typically represents expenditure, and the lower line represents tax revenue. The widening gap visualizes the fiscal deficit. The metaphor makes the structural imbalance immediately memorable.

Why It Matters

The chart is a good example of metaphor in data visualization. The underlying graphic is simple, but the metaphor gives the shape political and cultural meaning.

Design Notes

Make clear which line is expenditure and which is revenue.
Avoid letting the metaphor replace numerical explanation.
Use consistent scales so the gap is not exaggerated.
Explain the fiscal context behind the visual gap.

Summary

The crocodile-mouth chart shows how a familiar shape can become a public metaphor for data. It is a line chart, but its communicative force comes from the visual analogy.

Three-Dimensional Cylindrical Polar Scatter Graph

Sun, 12 Oct 2025 00:00:00 +0900

A three-dimensional cylindrical polar scatter graph uses cylindrical coordinates: angle, radius, and height. It visualizes three variables in a form related to a 3D extension of a polar plot. It is useful for multivariate data with directionality, periodicity, or radial structure.

How to Read It

The angle represents direction or phase, the radius represents distance or magnitude from the center, and height represents a third variable. Together they create a spatial distribution that can reveal directional and periodic patterns.

Use Cases

Meteorological direction and intensity data
Engineering measurements with angular components
Earth science and oceanographic data
Cyclical patterns with magnitude and height

Design Notes

Label angular units clearly.
Provide interaction or multiple views.
Avoid visual clutter from too many points.
Use color or transparency carefully.

Summary

This chart is specialized but useful when the data naturally fits cylindrical coordinates. It should be used when angle and radius are meaningful, not merely to make a chart look three-dimensional.

3D Bar Chart Map

Sat, 11 Oct 2025 00:00:00 +0900

A 3D bar chart map places vertical bars on geographic locations to show the distribution and comparison of quantitative data. It is often used for city- or region-level statistics such as population, GDP, sales, tourist volume, or disaster risk.

The height of each bar represents magnitude, while the map provides spatial context. In that sense, the technique combines the geographic intuition of a heat map with the direct quantity comparison of a bar chart.

How to Read It

Each bar on the map represents a specific place, such as a city, region, or grid cell.

Position: geographic coordinates, usually latitude and longitude
Height: the numeric value, such as population, sales, or volume
Color: an optional category, time period, or variable class

The 3D form can make large local differences easier to notice than they would be in a flat choropleth or heat map. In interactive maps, users can rotate, zoom, and change viewpoint to examine dense areas from different angles.

Background and Uses

3D bar chart maps became more common as GIS and WebGL-based visualization tools matured. Libraries such as Mapbox GL JS, deck.gl, and Kepler.gl provide extruded layers or column layers that make it relatively easy to render geographic data as interactive 3D columns.

Typical applications include:

comparing CO2 emissions by city
analyzing regional differences in real estate prices
showing GDP or trade volume by country
emphasizing flood depth, damage estimates, or other disaster-risk values

The form is also used in data art and editorial infographics because it can combine information density with a strong visual presence.

Summary

A 3D bar chart map is useful when readers need to understand both geographic distribution and quantitative contrast. Its visual impact is strong, and it works especially well in interactive exploration. At the same time, readability can suffer when the data is too dense, so bar scale, opacity, spacing, and camera angle must be tuned carefully. Used alongside a flat map, it can balance spatial context with numeric comparison.

3D Choropleth Map

Sat, 11 Oct 2025 00:00:00 +0900

A 3D choropleth map extends a choropleth map by using both color and height to represent values for geographic areas. Each region may be extruded into a prism whose height corresponds to a statistic such as population density or income.

Historical Background

Choropleth maps date back to early nineteenth-century statistical cartography. Three-dimensional extrusion became common with modern GIS, WebGL, and tools such as ArcGIS, Mapbox, deck.gl, and Kepler.gl.

Design Notes

Use with caution because perspective can mislead.
Avoid hiding small or low regions behind tall ones.
Provide interaction and 2D alternatives.
Clarify whether height, color, or both encode the value.

Summary

3D choropleth maps can be visually impactful, but they introduce occlusion and viewpoint dependency. They are best used when spatial extrusion supports exploration rather than replacing careful comparison.

3D Treemaps

Sat, 11 Oct 2025 00:00:00 +0900

3D treemaps extend ordinary treemaps into three-dimensional space. Rectangular areas represent hierarchical categories, while height encodes an additional quantitative variable. This extrusion-based form can show area and height together.

Historical Background

Treemaps were introduced by Ben Shneiderman in 1990 for visualizing file system disk usage. Later research explored 3D extensions as computer graphics tools improved.

Design Notes

Use 3D only when height adds meaningful information.
Watch for occlusion and perspective distortion.
Provide interaction or alternative 2D views.
Avoid using height and color redundantly without explanation.

Summary

3D treemaps can add another variable to hierarchical area charts, but they are harder to read precisely than 2D treemaps. They require careful camera and interaction design.

Adjacency Matrix

Sat, 11 Oct 2025 00:00:00 +0900

An adjacency matrix represents a network as a table. Nodes appear as both rows and columns, and each cell indicates whether a connection exists between the corresponding pair of nodes.

Purpose

The purpose is to show network structure without edge crossings. Matrices are especially useful for dense networks where node-link diagrams become unreadable.

Design Notes

Node ordering is crucial.
Use color intensity for edge weight.
Cluster related nodes to reveal block structure.
Provide labels or interaction for large matrices.

Summary

Adjacency matrices are precise and scalable network representations. They are less intuitive than node-link diagrams at first, but much clearer for dense graphs.

Arc Diagram

Sat, 11 Oct 2025 00:00:00 +0900

An arc diagram places nodes along a straight line and draws relationships as arcs above or below that line. It is a network visualization technique that can reveal patterns, symmetry, repetition, and long-range connections.

Design Notes

Node ordering strongly affects readability.
Use for moderate-sized networks.
Separate directions or categories with color or arc side when helpful.
Avoid dense networks with too many overlapping arcs.

Summary

Arc diagrams are simple network views where ordering is central. They work well when the sequence of nodes has meaning or when connection patterns are the main story.

BioFabric

Sat, 11 Oct 2025 00:00:00 +0900

BioFabric is a network visualization method that represents nodes as horizontal lines and edges as short vertical marks crossing those lines. This differs from conventional node-link diagrams and can make large networks more orderly.

Purpose

The purpose is to avoid node overlap and tangled links in large networks. By turning nodes into lines, BioFabric gives every node a visible row and every edge a position.

Design Notes

Ordering of nodes and edges is critical.
Use interaction to inspect details.
Explain the unfamiliar encoding to readers.
Consider it for large networks where node-link diagrams become unreadable.

Summary

BioFabric is a distinctive alternative to node-link diagrams. It can reveal structure in large networks by trading familiar geometry for a more scalable tabular-like layout.

Bivariate Choropleth Map

Sat, 11 Oct 2025 00:00:00 +0900

A bivariate choropleth map represents two variables on one map. Instead of using one color scale for one variable, it combines two color axes so each region’s color represents a paired value.

Purpose

The purpose is to show the spatial relationship between two variables, such as income and education, or vulnerability and exposure.

Design Notes

Use a simple 2x2 or 3x3 legend.
Choose variables that have a meaningful relationship.
Avoid too many color classes.
Make sure readers understand the mixed-color legend.

Summary

Bivariate choropleth maps are powerful for showing two-variable spatial patterns, but they require careful legends and restrained color design.

Box-and-Whisker Plot

Sat, 11 Oct 2025 00:00:00 +0900

A box-and-whisker plot summarizes a distribution using the minimum, first quartile, median, third quartile, and maximum, often with outliers shown as individual points. The box represents the interquartile range, or IQR.

Purpose

The purpose is to compare distributions compactly across groups. Box plots show center, spread, skew, and outliers without plotting every observation.

Design Notes

Explain the whisker rule when necessary.
Use alongside raw points for small datasets.
Do not hide important distribution shape when multimodality matters.
Consider violin plots when density shape is important.

Summary

Box plots are compact and robust distribution summaries. They are especially useful for comparing many groups.

Bullet Graph

Sat, 11 Oct 2025 00:00:00 +0900

A bullet graph is a compact chart for showing performance against a target. It extends a bar chart by combining an actual value, qualitative ranges such as poor, satisfactory, and good, and a target marker.

Historical Background

Bullet graphs were developed by Stephen Few as a more space-efficient and informative alternative to dashboard gauges.

Design Notes

Use a clear target marker.
Keep qualitative ranges subtle.
Avoid decorative gauge-like styling.
Use consistent scales when comparing multiple bullet graphs.

Summary

Bullet graphs are strong dashboard components because they show actual value, target, and performance context in a compact form.

Calendar Heatmap

Sat, 11 Oct 2025 00:00:00 +0900

A calendar heatmap visualizes date-based data using the structure of a calendar. Each day is represented by a cell, and color intensity shows the value for that day. It is useful for seeing daily activity, frequency, seasonality, and weekly patterns.

Use Cases

Contribution or activity logs
Daily sales or traffic
Habit tracking
Weather or environmental daily values

Design Notes

Make the calendar orientation clear.
Use a color scale suited to the data.
Label months and weekdays.
Consider missing days and holidays.

Summary

Calendar heatmaps are effective when the calendar structure itself matters. They reveal daily and seasonal patterns that ordinary line charts may hide.

Candlestick Chart

Sat, 11 Oct 2025 00:00:00 +0900

A candlestick chart visualizes open, high, low, and close prices over time. Each candlestick represents one period, such as a day or an hour. The body shows the difference between open and close, while the wicks show the high and low.

Purpose

Candlestick charts are widely used in financial markets because they summarize price movement within each time period compactly.

How to Read It

A rising period is often shown in white or green, and a falling period in black or red. Long bodies indicate large open-to-close movement; long wicks indicate wide intraperiod fluctuation.

Design Notes

Keep time intervals consistent.
Avoid overinterpreting individual shapes without context.
Use volume or moving averages only when they add analytical value.

Summary

Candlestick charts are standard for financial time series because they show four price values in one compact mark.

Chord Diagram

Sat, 11 Oct 2025 00:00:00 +0900

A chord diagram visualizes relationships or flows among categories arranged around a circle. Curved bands, or chords, connect categories, and their thickness or color represents strength or direction.

Use Cases

Migration flows
Trade between countries
Network relationships among categories
Genomic and biological relationships

Design Notes

Limit the number of categories.
Use ordering to reduce crossings.
Make direction clear if flows are directional.
Provide interaction for dense diagrams.

Summary

Chord diagrams are visually striking and useful for showing many-to-many relationships. They can become difficult to read when too many categories or chords are included.

Circle Voronoi Treemaps

Sat, 11 Oct 2025 00:00:00 +0900

Circle Voronoi treemaps visualize hierarchical data by partitioning an area into irregular polygons using Voronoi diagrams. Unlike rectangular treemaps, they use organic polygonal regions within a circular boundary, often producing a more natural-looking layout.

Historical Background

Voronoi treemaps were proposed in the early 2000s, especially through work by Michael Balzer and Oliver Deussen. They addressed limitations of rectangular treemaps, such as extreme aspect ratios and reduced readability.

Purpose

The purpose is to show hierarchical part-to-whole structure while allowing a more organic layout than rectangular subdivision.

Design Notes

Use labels carefully because polygon shapes vary.
Keep hierarchy depth manageable.
Use color to reinforce grouping.
Consider ordinary treemaps when precise area comparison is more important.

Summary

Circle Voronoi treemaps are visually distinctive hierarchical visualizations. They are useful when organic form and grouping matter, but they can be harder to compare precisely than rectangular treemaps.

Circular Chart

Sat, 11 Oct 2025 00:00:00 +0900

Circular chart is a broad term for visualizations arranged around a circular layout. Examples include pie charts, donut charts, radar charts, polar plots, radial bar charts, and circular time charts.

Purpose

Circular charts are often used for part-to-whole relationships, cyclical time, radial comparison, or emphasizing periodic patterns such as hours, months, seasons, and daily rhythms.

Design Notes

Use circular form when angle or cycle has meaning.
Avoid circular charts for ordinary ranking when bars would be clearer.
Label carefully because circular layouts can be hard to scan.
Be cautious with area and angle comparisons.

Summary

Circular charts can make cyclical structure visible, but they are not automatically better than rectangular charts. The circular layout should match the logic of the data.

Circular Layout

Sat, 11 Oct 2025 00:00:00 +0900

A circular layout places network nodes at equal intervals around a circle. It is useful when node position has no inherent spatial meaning and the goal is to show the overall structure simply.

Use Cases

Small network overviews
Comparing connection patterns around a fixed ordering
Teaching graph structure
Layouts supported by tools such as NetworkX

Design Notes

Choose node order carefully; it strongly affects edge crossings.
Avoid very dense networks.
Use color or grouping to support interpretation.
Consider chord diagrams when edge volume matters.

Summary

Circular layouts are simple and orderly. They work best when the circular arrangement itself supports reading the network.

Correlogram

Sat, 11 Oct 2025 00:00:00 +0900

A correlogram visualizes correlations among multiple variables in matrix form. Variables appear on both axes, and each cell encodes the correlation coefficient through color, size, shape, or numeric labels.

Purpose

The purpose is to give an overview of correlation structure: strong positive relationships, strong negative relationships, and weak or absent relationships.

Design Notes

Use a diverging color scale centered on zero.
Keep the variable order meaningful.
Show values when precision matters.
Remember that correlation does not imply causation.

Summary

Correlograms are compact summaries of relationships among many variables. They focus on strength and direction, while scatterplot matrices show the actual shape of relationships.

Density Plot

Sat, 11 Oct 2025 00:00:00 +0900

A density plot shows the distribution of a continuous variable as a smooth curve. Instead of using histogram bins, it estimates a probability density function, often with kernel density estimation.

Purpose

Density plots are useful for comparing distribution shapes, including peaks, spread, skewness, and overlap between groups.

Design Notes

Choose bandwidth carefully.
Explain that the curve is an estimate.
Use transparency when comparing multiple groups.
Consider histograms when raw counts are important.

Summary

Density plots provide a smooth view of distribution. They are useful for comparison, but their appearance depends on the estimation method and bandwidth.

Diverging Stacked Bar Chart

Sat, 11 Oct 2025 00:00:00 +0900

A diverging stacked bar chart compares opposing response categories around a central baseline. It is commonly used for Likert-scale survey results, where agreement and disagreement are placed on opposite sides of zero.

Purpose

The purpose is to show the balance between positive and negative responses while preserving category composition.

Design Notes

Place neutral responses carefully, either centered or separated.
Use consistent colors for negative and positive categories.
Sort items by net agreement or another meaningful metric.
Avoid too many response levels.

Summary

Diverging stacked bar charts are strong for survey and sentiment data because they make polarity visible. They require careful treatment of neutral categories and color order.

Dot Density Map

Sat, 11 Oct 2025 00:00:00 +0900

A dot density map represents geographic distribution through the density of dots. Each dot stands for a fixed quantity, such as 100 people or one household, and dots are placed within areas to show regional patterns.

Purpose

The purpose is to show spatial concentration and spread while avoiding the area bias of choropleth maps. Dense dot clusters indicate high values or high concentration.

Design Notes

State what one dot represents.
Avoid implying exact locations when dots are randomly placed within areas.
Use small dots and appropriate transparency.
Consider dasymetric placement when internal distribution is known.

Summary

Dot density maps are useful for showing population and other distributed quantities. They communicate density intuitively, but readers must understand that dots may represent quantities, not exact locations.

Dot Map

Sat, 11 Oct 2025 00:00:00 +0900

A dot map plots individual observations as dots at their geographic locations. One dot represents one event, object, or observation. It is also called a dot distribution map, and is useful for showing the geographic distribution of precise locations such as earthquakes, stores, or incidents.

Historical Background

Dot maps have roots in nineteenth-century cartography. John Snow’s 1854 cholera map in London is a famous early example of plotting individual cases geographically.

Design Notes

Use dot maps when locations themselves matter.
Avoid overplotting by using transparency or clustering.
Do not confuse dot maps with dot density maps, where one dot may represent many units.
Provide enough basemap context for interpretation.

Summary

Dot maps are direct and powerful when the data consists of individual locations. They show where observations occur without aggregating them into regions.

Dot Plot

Sat, 11 Oct 2025 00:00:00 +0900

A dot plot represents each data point as a dot. Dots are placed along an axis, and repeated values can be stacked to show frequency. It is useful for small to medium datasets where the individual observations should remain visible.

Purpose

Dot plots show distribution without hiding individual values. They can reveal clusters, gaps, outliers, and repeated values more directly than summary-only charts.

Design Notes

Use for small or moderate datasets.
Stack or jitter dots to reduce overlap.
Keep axes simple.
Consider histograms or density plots for large datasets.

Summary

Dot plots are simple, transparent distribution charts. They are especially good when readers should see both individual values and overall pattern.

Force-Directed Diagram

Sat, 11 Oct 2025 00:00:00 +0900

A force-directed diagram is a network layout method that positions nodes and edges through simulated physical forces. Nodes repel one another, while edges act like springs that pull connected nodes together. The layout settles into a balanced configuration that can reveal clusters and relationships.

Use Cases

Social networks
Web link structures
Biological interaction networks
Knowledge graphs
Organizational relationship maps

Design Notes

Use interaction for large networks.
Avoid treating distance as exact unless the layout defines it.
Label only important nodes in dense diagrams.
Consider adjacency matrices or hive plots when comparison and structure matter more than organic layout.

Summary

Force-directed diagrams are intuitive for exploring network structure, especially clusters and connected groups. Their apparent geometry is generated by layout rules, so it should be interpreted carefully.

Funnel Chart

Sat, 11 Oct 2025 00:00:00 +0900

A funnel chart visualizes the decrease in quantity across stages of a process. Its width narrows from top to bottom, resembling a funnel. It is common in marketing and sales analysis, such as website visits, product views, cart additions, and completed purchases.

Historical Background

The funnel concept in marketing traces back to the AIDA model proposed by E. St. Elmo Lewis in 1898: Attention, Interest, Desire, and Action. It describes how people move through stages before purchase.

How to Read It

Each stage shows the number or percentage remaining. The drop between stages indicates loss or conversion. The most important insight is often where the largest drop-off occurs.

Design Notes

Use consistent stage definitions.
Show both counts and conversion rates when useful.
Avoid 3D funnel effects.
Consider bar charts when exact comparison is more important than the funnel metaphor.

Summary

Funnel charts are useful for communicating staged drop-off in processes. They are most effective when the process truly narrows from one stage to the next.

Gantt Chart

Sat, 11 Oct 2025 00:00:00 +0900

A Gantt chart visualizes project tasks, durations, order, and dependencies with horizontal bars on a time axis. Each bar shows the start and end of a task, making the overall schedule easy to understand.

Historical Background

Gantt charts were developed in the 1910s by Henry Laurence Gantt, who was influenced by scientific management. They were used in production planning and later became standard tools for project management.

Data Structure

Data	Role
Task	One row
Start date	Bar beginning
End date or duration	Bar length
Dependency	Optional relationship
Status	Optional color or progress

Design Notes

Show dependencies when they affect schedule.
Keep time scale readable.
Use color for status or responsibility.
Avoid overloading the chart with too many small tasks.

Summary

Gantt charts are practical tools for planning and communicating schedules. They are strongest when the key question is “what happens when?”

Gauge Chart

Sat, 11 Oct 2025 00:00:00 +0900

A gauge chart shows a single value on a circular or semicircular scale, often with a needle like a speedometer. It is commonly used to show where a current value sits relative to a target, threshold, or acceptable range.

Use Cases

Dashboard KPIs
Operational monitoring
Performance against target
Status displays

Design Notes

Use gauges for quick status, not detailed analysis.
Show thresholds clearly.
Avoid wasting too much dashboard space.
Consider bullet charts or bar charts when comparison matters.

Summary

Gauge charts are familiar and easy to scan, but they are space-inefficient and limited to simple status messages. Use them when the metaphor of a meter helps the audience.

Geography-Based Connection

Sat, 11 Oct 2025 00:00:00 +0900

A geography-based connection map visualizes network relationships on top of geographic space. Nodes such as cities or locations are placed on a map, and edges connect them to show movement, communication, logistics, or other relationships.

Historical Background

The form is related to nineteenth-century flow maps and later airline route maps, communication network maps, and logistics maps. Modern GIS and network tools make it easy to overlay network structure on digital maps.

Use Cases

Air routes
Internet and communication cables
Logistics and supply chains
Migration or commuting connections
Urban and regional networks

Design Notes

Avoid too many overlapping edges.
Use bundling or filtering for dense networks.
Clarify whether line width represents volume.
Preserve geographic context without overwhelming the network.

Summary

Geography-based connection maps are useful when the location of relationships matters. They combine network structure with geographic context.

Histogram

Sat, 11 Oct 2025 00:00:00 +0900

A histogram visualizes the distribution of numeric data. The value range is divided into continuous intervals, or bins, and each bar shows how many observations fall into that bin. Because histograms show continuous distributions, bars usually touch one another.

Historical Background

Karl Pearson introduced the term “histogram” in 1895. The visual idea belongs to the broader history of statistical graphics, but Pearson helped establish the term and its theoretical role in statistics.

How to Read It

Look at the shape of the bars: peaks, spread, skewness, gaps, and outliers. The bin width affects the appearance, so a histogram should be interpreted with that choice in mind.

Design Notes

Choose bin width carefully.
Do not use gaps between bars for continuous data.
Label units and axes.
Consider density plots for smoother distribution views.

Summary

Histograms are fundamental for understanding numeric distributions. They are simple, but binning choices strongly affect what readers see.

Hive Plot

Sat, 11 Oct 2025 00:00:00 +0900

A hive plot visualizes network data by placing nodes on predefined axes according to explicit rules. Unlike force-directed diagrams, it avoids random-looking layouts and makes network structure more systematic and comparable.

Purpose

The purpose is to make complex networks more interpretable by controlling node placement. Nodes are assigned to axes by category, role, or metric, and edges show relationships among them.

Design Notes

Define axis rules clearly.
Use consistent node ordering along axes.
Avoid too many edge crossings.
Use color to reinforce categories.

Summary

Hive plots are useful when a network should be read through defined structural dimensions rather than through a force-directed layout. Their clarity depends on meaningful axis design.

Hyperbolic Tree

Sat, 11 Oct 2025 00:00:00 +0900

A hyperbolic tree visualizes hierarchical data in a circular hyperbolic space. Elements near the center are enlarged, while elements toward the edge are compressed. This makes it possible to see both a focus area and surrounding context, making it a classic focus-plus-context technique.

Purpose

The purpose is to navigate large hierarchies without losing the broader structure. Users can bring a node into focus while keeping related branches visible around it.

Use Cases

Large file systems
Website maps
Taxonomies
Knowledge graphs with tree-like structure

Design Notes

Use interaction; static hyperbolic trees are harder to read.
Keep labels legible near the focus.
Provide orientation cues so users do not get lost.

Summary

Hyperbolic trees are powerful for interactive navigation of large hierarchies. Their strength is combining detail and context in one view.

Isotype Chart

Sat, 11 Oct 2025 00:00:00 +0900

An Isotype chart uses pictograms to compare quantities. Repeating icons of the same size allows readers to count or estimate differences intuitively. The system was developed in the 1920s by Otto Neurath and designer Gerd Arntz.

Historical Background

ISOTYPE, the International System Of Typographic Picture Education, originated in Vienna. Neurath wanted a visual language for communicating social and economic data to the general public. Arntz’s pictograms gave the system a clear and consistent graphic form.

Design Notes

Use repeated units rather than scaling pictogram area.
Make the value represented by one icon clear.
Keep pictograms simple and consistent.
Avoid partial icons when they confuse counting.

Summary

Isotype charts are effective for public communication because they turn quantities into countable visual units. They are strongest when the goal is accessibility and memorability rather than exact precision.

Radar Chart

Sat, 11 Oct 2025 00:00:00 +0900

A radar chart compares multiple variables by arranging axes radially and connecting values into a polygon. It is also called a spider chart or web chart. It is often used to compare attributes of products, people, organizations, or performance profiles.

Historical Background

Early forms appeared in nineteenth-century statistical graphics and were later used in military, industrial, and educational assessment. Software such as Excel, Tableau, matplotlib, and Plotly made radar charts easy to create.

Design Notes

Use a small number of variables.
Keep axis scales consistent and meaningful.
Avoid comparing too many polygons.
Consider small multiples or bar charts for precise comparison.

Summary

Radar charts are useful for profile-like comparisons, but they can be hard to read precisely. They work best when the overall shape is more important than exact values.

Radial Bar Chart

Sat, 11 Oct 2025 00:00:00 +0900

A radial bar chart places bars around a circle using polar coordinates. Each bar extends outward from the center, and its length represents a value. It is also called a circular bar chart or star chart.

Use Cases

Cyclical time data such as months, seasons, or hours
Decorative summaries for dashboards or reports
Comparing values in a circular ordering

Design Notes

Use with caution when precise comparison matters.
Avoid too many bars.
Label angles and categories clearly.
Consider a normal bar chart when cyclic structure is not meaningful.

Summary

Radial bar charts can be effective for cyclical data, but they are less precise than ordinary bar charts. The circular form should serve the data, not only decoration.

Radial Tree

Sat, 11 Oct 2025 00:00:00 +0900

A radial tree lays out a hierarchy in a circular form. The root node is placed at the center, and child nodes radiate outward by level. This makes hierarchical depth and parent-child relationships visible in a compact layout.

Historical Background

Circular tree layouts appear in nineteenth-century classification and phylogenetic diagrams, but computational radial tree layouts developed in information visualization research in the late twentieth century.

Use Cases

Taxonomies
Organization structures
File systems
Phylogenetic trees
Information architecture maps

Design Notes

Use radial trees when compact overview matters.
Keep labels legible around the circle.
Avoid very deep hierarchies without interaction.
Consider horizontal trees when labels are long.

Summary

Radial trees are compact hierarchy diagrams. They provide a strong overview, but label readability and dense branches require care.

Ridgeline Plot

Sat, 11 Oct 2025 00:00:00 +0900

A ridgeline plot stacks multiple distributions, usually smooth density curves, with slight vertical offsets. It is useful for comparing how distributions change across time or categories. The name comes from the appearance of mountain ridge lines.

Use Cases

Temperature distributions over months
Income or demographic distributions by group
Music waveform-like patterns
Comparing many related density curves

How to Read It

Each ridge shows one distribution. Peaks indicate where values are concentrated. Comparing ridge positions and shapes reveals shifts, spread, and changes in modality.

Design Notes

Avoid excessive overlap that hides lower ridges.
Use consistent scales.
Label categories clearly.
Consider violin plots or small multiples when exact comparison matters.

Summary

Ridgeline plots are effective for comparing many distributions compactly. They emphasize shape and change more than exact values.

Space-Time Cube

Sat, 11 Oct 2025 00:00:00 +0900

A space-time cube integrates time and geographic space in a three-dimensional visual form. The X and Y axes represent location, while the Z-axis represents time. This makes it possible to see spatial and temporal change together.

Historical Background

The concept is rooted in time geography, developed by Swedish geographer Torsten Hagerstrand in the 1960s. His ideas about space-time prisms showed how human activity is constrained by both space and time. Later GIS and 3D visualization tools made space-time cubes practical.

Data Structure

Data	Role
X/Y location	Geographic position
Time	Vertical axis
Event or path	Point, line, or trajectory
Attribute	Color, size, or annotation

Design Notes

Use interaction or multiple views for readability.
Clarify the time direction.
Avoid clutter from too many trajectories.
Provide 2D map and timeline references when needed.

Summary

Space-time cubes are useful when spatial and temporal patterns must be understood together. They are powerful but often require interaction to read well.

Spiral Chart

Sat, 11 Oct 2025 00:00:00 +0900

A spiral chart represents time or cyclical change along a spiral path. Unlike a circular chart that completes in one loop, a spiral can wind continuously through multiple cycles, making seasonal and periodic patterns easier to compare across years or periods.

Historical Background

Spiral forms in information graphics date back to nineteenth-century statistical diagrams. They are conceptually related to polar charts and cyclical time representations.

Use Cases

Seasonal time series
Daily, weekly, or yearly cycles
Long time series with repeated periods
Periodic environmental or social data

Design Notes

Make the time direction clear.
Avoid using spiral form when linear time comparison is simpler.
Label cycles or reference periods.
Use color carefully to avoid confusing position and value.

Summary

Spiral charts are useful when periodicity is central to the story. They make repeated cycles visible, but they can be harder to read precisely than linear time-series charts.

Stem-and-Leaf Plot

Sat, 11 Oct 2025 00:00:00 +0900

A stem-and-leaf plot splits numeric values into stems and leaves, preserving individual observations while showing the shape of the distribution. It resembles a histogram, but the original data values remain visible.

Historical Background

Stem-and-leaf plots were popularized by John W. Tukey as part of exploratory data analysis in the 1970s. They were designed as a quick, hand-drawn way to inspect distributions before computers were widely available.

How to Read It

The stem contains the leading digits, and each leaf contains the final digit or digits. The row length shows frequency, while the leaves preserve individual values.

Design Notes

Use for small to medium datasets.
Choose stem width carefully.
Sort leaves for readability.
Use histograms for larger datasets.

Summary

Stem-and-leaf plots are useful for teaching and exploratory analysis because they reveal distribution while keeping the raw values visible.

Strip Chart

Sat, 11 Oct 2025 00:00:00 +0900

A strip chart records or displays data continuously over time, often as a line updated in real time. The term also refers to charts that show individual data points as a continuous strip across time or categories. The common idea is to make ongoing change visible.

Historical Background

Strip charts originated in physical strip chart recorders used for industrial measurement, medical monitoring, and experiments. A moving paper strip recorded changing values as a continuous trace. The idea later carried over into digital monitoring displays and statistical plots.

Use Cases

Industrial sensor monitoring
Medical and laboratory measurement
Real-time system dashboards
Time-series inspection

Design Notes

Keep time scale and sampling rate clear.
Use alerts or bands for important thresholds.
Avoid overplotting too many signals.
Preserve enough history for context.

Summary

Strip charts are practical tools for monitoring change as it happens. Their value comes from continuity and immediacy.

Symbol Chart

Sat, 11 Oct 2025 00:00:00 +0900

A symbol chart uses geometric symbols such as circles or squares, varying their size or color to represent quantities and categories. Symbols may be placed on ordinary axes or on maps. Bubble charts and proportional symbol maps are closely related forms.

Historical Background

Symbol-based visualization has roots in nineteenth-century cartography. Proportional symbols appeared on railway and thematic maps, and Charles Joseph Minard developed influential uses of symbol size for flows and quantities. Later work on area perception, including James Flannery’s scaling research, improved how proportional symbols are designed.

How to Read It

Read position first, then symbol size and color. Size usually represents magnitude; color may represent category or another quantitative value.

Design Notes

Scale area, not diameter, when symbol size represents value.
Provide a size legend.
Avoid excessive overlap.
Use color only when it adds meaning.

Summary

Symbol charts are flexible because they combine position, size, and color. They are useful for showing magnitude across locations or categories, but scaling and overlap require careful handling.

Time-Series Linear Dendrogram

Sat, 11 Oct 2025 00:00:00 +0900

A time-series linear dendrogram unfolds hierarchical relationships along a time axis. Whereas ordinary dendrograms mainly show spatial hierarchy, the linear time-series form shows when and how branches diverge or emerge.

Historical Background

Dendrograms developed from biological classification and later became standard in hierarchical clustering. Time-aware linear dendrograms are especially important in phylogenetics, where mutations and lineages must be understood over time. Platforms such as Nextstrain made this form widely visible during the COVID-19 pandemic.

Data Structure

Data	Role
Node	Ancestor, group, or cluster
Branch	Relationship or divergence
Time	Position along the timeline
Metadata	Optional color or annotation

Purpose

The purpose is to show hierarchical branching together with temporal sequence. It answers not only what is related to what, but when divergence occurred.

Design Notes

Make the time axis clear.
Use color for meaningful lineage or category metadata.
Avoid overloading the chart with too many labels.
Provide interaction for dense trees.

Summary

Time-series linear dendrograms are useful when hierarchy and time are inseparable. They are especially powerful for evolution, lineage tracking, and historical branching processes.

Vector Field Map

Sat, 11 Oct 2025 00:00:00 +0900

A vector field map shows direction and magnitude at spatial locations using arrows or glyphs. It is used for data such as wind, ocean currents, electric fields, and magnetic fields. Arrow angle represents direction, while length or color represents strength.

Historical Background

Vector field visualization developed with physics and fluid dynamics. Michael Faraday’s field-line diagrams and James Clerk Maxwell’s mathematical work on fields helped establish visual and mathematical thinking about directional forces. Modern weather and ocean maps now routinely use vector field visualization.

Data Structure

Data	Role
Location	Point where vector is measured
Direction	Arrow angle
Magnitude	Arrow length or color
Optional time	Animation or temporal comparison

Design Notes

Avoid drawing too many arrows.
Use consistent scaling for magnitude.
Consider streamlines for dense flow.
Use animation carefully for time-varying fields.
Include a legend for arrow length or color.

Summary

Vector field maps are effective for directional spatial data. They help readers see flow, force, or movement patterns, but density and scaling must be managed carefully.

Vertical Trees and Horizontal Trees

Sat, 11 Oct 2025 00:00:00 +0900

Vertical and horizontal trees are basic layouts for visualizing hierarchical structures with nodes and branches. A vertical tree places the root at the top and expands downward. A horizontal tree expands from left to right or right to left. Both are used for organization charts, taxonomies, phylogenetic trees, file structures, and other hierarchies.

Historical Background

Tree diagrams have roots in ancient classification diagrams, medieval genealogies, and scientific evolutionary trees. In computer science, trees became a fundamental data structure, and visualization libraries now generate many layout variants automatically.

Data Structure

Data	Role
Node	Entity in the hierarchy
Parent-child relationship	Branch structure
Root	Top or starting node
Depth	Level in the hierarchy

Design Notes

Use vertical trees for top-down hierarchy.
Use horizontal trees when labels are long.
Keep spacing consistent.
Collapse or filter deep hierarchies.
Consider radial trees for compact overviews.

Summary

Vertical and horizontal trees are simple, readable ways to show hierarchy. The best orientation depends on label length, reading direction, and available space.

Violin Plot

Sat, 11 Oct 2025 00:00:00 +0900

A violin plot visualizes both the distribution shape and summary statistics of data. It builds on the box plot but adds a smooth density shape, usually based on kernel density estimation. The result often resembles a violin, giving the chart its name.

Purpose

The purpose is to show not only median and spread, but also the shape of the distribution: modes, skewness, and density.

How to Read It

The width of the violin indicates density. Wider areas contain more observations. Internal marks may show the median, quartiles, or individual data points depending on the design.

Use Cases

Comparing distributions across categories
Showing multimodal data
Supplementing or replacing box plots
Exploratory data analysis

Design Notes

Explain whether the plot shows KDE, raw points, or both.
Be careful with small sample sizes.
Use consistent bandwidth settings across groups.
Consider box plots when summary statistics are enough.

Summary

Violin plots are useful when distribution shape matters. They reveal more than box plots, but their density estimates require careful interpretation.

Waterfall Chart

Sat, 11 Oct 2025 00:00:00 +0900

A waterfall chart visualizes incremental increases and decreases from an initial value to a final value. It separates positive, negative, and total components so readers can understand what caused the overall change.

The name comes from the way bars appear to step up and down like a waterfall.

Purpose

Waterfall charts are especially common in finance, where they explain how revenue, costs, adjustments, and other components lead to profit or loss.

How to Read It

Start with the initial value, then follow each bar in sequence. Upward bars add to the total, downward bars subtract from it, and the final bar shows the resulting value.

Design Notes

Use color to distinguish increases, decreases, and totals.
Keep the sequence logical.
Label major components directly.
Avoid too many small steps.

Summary

Waterfall charts are useful for explaining change as a sequence of contributions. They make bridge analysis easier to understand than a single before-and-after comparison.

Yield Curve

Sat, 11 Oct 2025 00:00:00 +0900

A yield curve connects the yields of bonds with different maturities. The X-axis usually shows maturity, and the Y-axis shows yield. It is an important financial chart for reading market expectations and the economic outlook.

Historical Background

The concept developed with bond markets and theories of the term structure of interest rates. Economists and investors use yield curves to understand expectations about inflation, monetary policy, and future growth.

How to Read It

A normal upward-sloping curve means longer maturities have higher yields. A flat curve suggests uncertainty or transition. An inverted curve, where short-term yields exceed long-term yields, is often watched as a recession signal.

Design Notes

Label maturities clearly.
Use consistent dates when comparing curves.
Avoid too many curves in one chart.
Consider animation or small multiples for historical change.

Summary

The yield curve is a compact view of interest-rate expectations across maturities. Its shape is often more important than any single point.

Circle Packing

Thu, 09 Oct 2025 00:00:00 +0900

Circle packing visualizes whole-part and parent-child relationships by packing circles inside other circles. It is a hierarchical visualization technique: each node is drawn as a circle, and parent nodes contain their children.

Compared with a treemap, which divides rectangular space, circle packing is less precise for area comparison but often more intuitive for showing containment and hierarchy depth.

Background

The mathematical idea of packing circles is old and has been studied as the circle packing problem since the nineteenth century. In information visualization, circle packing became widely used in the late 1990s and 2000s, especially through tools such as Protovis and D3.js. D3 provides a standard implementation as d3.pack().

Data Structure

Circle packing uses hierarchical data.

Field	Meaning
name	Node label
value	Numeric size for the node
children	Child nodes

Example:

{
  "name": "root",
  "children": [
    {"name": "A", "value": 10},
    {"name": "B", "children": [
      {"name": "B1", "value": 5},
      {"name": "B2", "value": 8}
    ]}
  ]
}

Purpose

The main purpose is to show which elements belong to which groups, and roughly how large each group or leaf node is. It is visually softer than a treemap and often works well when structural hierarchy matters more than exact numeric comparison.

Use Cases

file-system structure
biological taxonomy
corporate or industry hierarchy
regional population composition
topic clusters and nested categories

How to Read It

Element	Meaning
Circle size	Relative value or magnitude
Parent and child circles	Hierarchical containment
Color	Category, group, or level
Position	Layout optimized for readability; not usually a measured axis

Design Notes

Area comparison with circles is difficult, so labels and tooltips help.
Deep hierarchies often need interactive zooming.
Use color consistently by level or parent group.
Leave enough padding between circles so boundaries remain legible.

Alternatives

Method	Feature	Best when
Treemap	Rectangular area optimization	Accurate area comparison matters
Sunburst chart	Radial hierarchy	Hierarchy depth matters
Icicle chart	Stacked hierarchy	Labels need more room

Summary

Circle packing is a visually compelling way to show hierarchical structure. It is not the strongest option for exact value comparison, but it is effective when the goal is to reveal the shape of an information hierarchy.

References

Time-Distance Diagram

Thu, 09 Oct 2025 00:00:00 +0900

A time-distance diagram visualizes the operation of moving vehicles, such as trains or buses, by placing time on one axis and distance or station position on the other. Each train or vehicle is drawn as a slanted line, making schedules, passing points, stops, and service patterns easy to understand at a glance.

The method is widely used in railway planning and operations, where it is often called a train graph or timetable diagram.

Historical Background

Time-distance diagrams emerged alongside the expansion of railway networks in the nineteenth century. Similar operating diagrams appeared in British railway practice in the 1870s, and in Japan the format became a standard tool for timetable design and dispatching during the Meiji period.

Even today, train-control centers rely on this type of diagram to monitor services and manage disruptions.

Data Structure

Data item	Description
Station or stop name	Reference points along the distance axis
Distance between stations	Basis for the vertical scale
Departure and arrival times	Time values plotted on the horizontal axis
Train number or service type	Used for line style, color, and labels

Each service is drawn as a line segment through time and space. Depending on local convention, inbound and outbound trains may slope in opposite directions.

Purpose

The main purpose is to optimize service planning and operations. A time-distance diagram helps planners see whether the whole timetable is coherent, where trains meet or overtake one another, how transfers work, and where congestion or delays may occur.

Use Cases

railway dispatching and timetable design
bus, ferry, and public-transport scheduling
recovery planning after service disruptions
transport simulation in research and education
logistics and autonomous-vehicle movement analysis

How to Read It

Horizontal axis: time
Time progresses from left to right.
Vertical axis: distance or station sequence
Stations or locations are arranged by physical distance or by service order.
Slanted lines: vehicle movement
The slope represents speed. A flatter line indicates faster movement; a steeper line indicates slower movement.
Intersections: meets, passes, or transfer opportunities
Where lines cross, vehicles are at the same position at the same time.
Horizontal segments: stops or waiting time
Longer horizontal segments indicate longer dwell or waiting periods.

Design Notes

Use a true distance scale when speed comparison matters.
Use color or line style for service types such as local, express, or limited express.
Keep the time grid regular so departure and arrival times can be read precisely.
Use interaction or scrolling for dense digital timetables.

Alternatives

Gantt chart: useful for general schedules and tasks.
Sparkline time series: useful for frequency or headway trends.
Network diagram: useful when topology matters more than timing.

Summary

A time-distance diagram is a compact way to show movement through both time and space. Its structure is simple, but it can reveal speed, delay, connection, passing, and operational rhythm in a single view.

References

Euler Diagram

Wed, 08 Oct 2025 00:00:00 +0900

An Euler diagram visualizes inclusion and overlap among sets. Unlike a Venn diagram, which draws every logical combination, an Euler diagram shows only the relationships that actually exist. This makes it useful for explaining real categories, concepts, and classifications clearly.

Historical Background

Euler diagrams are named after the eighteenth-century Swiss mathematician Leonhard Euler, who used circles to explain logical relationships and syllogisms. Later, John Venn extended this tradition into diagrams that include all logical possibilities, producing what we now call Venn diagrams.

Euler diagrams remain more concise when the goal is to communicate actual relationships rather than theoretical completeness.

This image is an example of Eulerian circles associated with Euler’s Lettres a une princesse d’Allemagne. The circles visually express logical relationships: intersection means shared members, containment means “all A are B,” and separation means no overlap.

Data Structure

Element	Description
Set	A group represented by a circle or closed curve
Element	A data point belonging to a set
Intersection	Elements shared by two or more sets
Complement	Elements outside a given set

An Euler diagram omits empty intersections when they do not exist in the data.

Purpose

The main purpose is to show real set relationships simply. By leaving out impossible or empty regions, Euler diagrams can be easier to read than Venn diagrams for practical classification and presentation.

Use Cases

explaining conceptual or category relationships
showing overlap between tags or labels
teaching set inclusion and subset relationships
visualizing target groups in UX or marketing
comparing product features or audience segments

How to Read It

Each circle or closed curve represents a set.
Overlap indicates shared elements.
Containment indicates a subset relationship, such as A ⊆ B.
Missing intersections may simply mean that no such relationship exists.

Design Notes

Keep the number of sets small, ideally three or four.
Use transparent colors if regions overlap.
Place labels inside or near the relevant region.
Avoid shapes that make containment or intersection ambiguous.

Alternatives

Method	Feature	Use case
Venn diagram	Shows all combinations	Logic and education
Network diagram	Nodes and links	Complex relationships
Treemap	Hierarchical area	Category proportions
Matrix	Tabular intersections	Many combinations

Euler Diagram vs. Venn Diagram

Venn diagram: shows all 2^n logical combinations, including empty regions.
Euler diagram: shows only relationships that actually occur.

This makes Euler diagrams especially useful for realistic classification, such as showing that cats are mammals and mammals are animals.

Summary

Euler diagrams are practical tools for showing real-world relationships among sets. They give up the theoretical completeness of Venn diagrams in exchange for clarity and economy.

References

Venn Diagram

Wed, 08 Oct 2025 00:00:00 +0900

A Venn diagram visualizes relationships between sets. Circles or closed curves represent sets, and their overlaps represent shared elements. Venn diagrams are used in logic, probability, set theory, education, data analysis, and business communication to make inclusion and exclusion relationships easy to understand.

Historical Background

The Venn diagram was formalized by the British logician John Venn (1834-1923) in 1880. His paper “On the Diagrammatic and Mechanical Representation of Propositions and Reasonings” presented a systematic way to show set relationships.

Earlier diagrams by Leonhard Euler also represented set relationships, but Euler diagrams show only relationships that actually exist. Venn diagrams differ by including all logical possibilities, or 2^n regions for n sets.

Data Structure

Element	Meaning
Set	A group represented by a circle or closed curve
Element	An item belonging to one or more sets
Intersection	Elements shared by multiple sets
Exclusive region	Elements belonging only to one set

Purpose

The purpose of a Venn diagram is to show similarities, differences, intersections, and exclusions among a small number of sets. It is especially effective for two or three sets and is widely used as a thinking and teaching tool.

Use Cases

logic and set theory
probability explanation
customer or audience segmentation
comparison of categories in data analysis
classroom exercises for similarities and differences

How to Read It

Each circle is a set.
Overlapping areas show shared elements.
Non-overlapping areas show elements unique to one set.
In a three-set Venn diagram, the central overlap contains elements shared by all three sets.

Symbol	Meaning	Position
`A`, `B`	Sets	Circles or closed regions
`A ∪ B`	Union	All parts of both circles
`A ∩ B`	Intersection	Overlapping area
`A - B`	Difference	Area in A but not B
`A^c`	Complement	Everything outside A

Design Notes

Use color and transparency so overlaps remain visible.
Place labels carefully to avoid obscuring intersections.
Avoid using Venn diagrams for precise quantitative area comparison unless the geometry is explicitly constructed for that purpose.
For many sets, consider Euler diagrams or UpSet plots.

Venn Diagram vs. Euler Diagram

Item	Venn diagram	Euler diagram
Main idea	Shows all logical possibilities	Shows only actual relationships
Regions	`2^n` possible regions	Only needed regions
Best for	Logic and theoretical explanation	Real-world classification
Shape	Usually circles or closed curves	Any clear closed curve

Summary

The Venn diagram is a foundational visual method for explaining relationships between sets and concepts. Its simple shape carries a rigorous idea: all possible logical combinations are represented, whether or not every region contains elements.

References

Waffle Chart

Sat, 04 Oct 2025 00:00:00 +0900

A waffle chart represents a whole as a grid of small cells, often 100 cells in a 10 by 10 layout. Each cell is colored by category, making proportions easy to understand as countable blocks.

Like pie charts and bar charts, waffle charts show part-to-whole relationships, but they have a more iconic and infographic-like appearance.

Background

The idea is related to ISOTYPE, the visual language promoted by Otto Neurath. ISOTYPE emphasized showing quantity through repeated symbols rather than changing the size of symbols.

Waffle charts became especially popular in data journalism and infographics after the 2010s as an alternative to pie charts. They are now supported or easily built in Python, R, Tableau, D3.js, and many presentation tools.

Data Structure

Category	Percentage
A	40
B	35
C	25

These values are assigned to cells in a fixed grid. In a 100-cell chart, each cell usually represents 1%.

Purpose

The goal is to show how a whole is divided into parts. Because the viewer can mentally count blocks, waffle charts often feel more concrete than angle-based pie charts.

Use Cases

workforce composition
gender ratio or demographic shares
budget and expense composition
survey responses
recycling rate or renewable-energy share

How to Read It

The full grid represents 100%.
Filled cells represent a category’s share.
Color connects cells to categories through a legend.
Small multiples allow fair comparison across groups.
Filling direction should remain consistent.

Design Notes

Use clear color contrast.
Choose an appropriate grid size, such as 10x10 or 20x5.
Avoid high-precision claims when rounding is involved.
Keep legends and labels clear.

Alternatives

Chart	Comparison
Pie chart	Uses angle and area to show part-to-whole composition
Bar chart	Better for exact quantitative comparison
Treemap	Shows part-to-whole relationships and hierarchy by area

Summary

Waffle charts turn proportions into visible, countable units. They are useful for communicating approximate shares, progress, and composition, especially in educational and editorial contexts.

References

Sunburst Chart

Thu, 02 Oct 2025 00:00:00 +0900

A sunburst chart visualizes hierarchical data in a circular layout. The root sits at the center, and each outer ring represents a deeper level in the hierarchy. It can be understood as a radial treemap or a multi-level pie chart.

Background

Sunburst charts developed from the broader history of compact hierarchy visualization. Their lineage includes early concentric classification diagrams, financial diagrams, and information-visualization research.

Important historical threads include:

Late nineteenth century: Lawrence W. Fike’s zoological concentric chart, which arranged biological taxonomy in rings.
1921: a “sunburst diagram” in Mechanical Engineering showing U.S. federal spending.
Around 2000: John Stasko and Georgia Tech research introduced radial space-filling views for exploring file systems and other hierarchies.

The form became widely used with web visualization, especially through D3.js and Mike Bostock’s partition layout.

Data Structure

Sunburst charts require tree-structured data.

Element	Meaning
Root node	Top-level category
Child node	Subcategory
Arc	Visual representation of a node
Angle or area	Quantitative value

Purpose

The goal is to show how parts belong to a whole across multiple levels. Sunburst charts are useful for product categories, file systems, organizational structures, website navigation, and hierarchical budgets.

Strengths

Shows hierarchy depth through radius.
Provides a compact overview.
Makes parent-child relationships visually intuitive.
Works well with interaction, such as click-to-zoom.

Limitations

Deep hierarchies make the outer rings crowded.
Exact comparison is difficult because values are encoded by angles and arcs.
Node ordering can affect interpretation.
User studies do not always favor sunburst charts for precise tasks.

How to Read It

The center is the highest-level category. Moving outward means moving down the hierarchy. Arc length usually represents value, and color separates categories or branches.

Design Notes

Use related colors for parent and child nodes.
Use tooltips or hover labels for outer rings.
Keep the hierarchy shallow when possible.
Use interaction when detailed comparison matters.

Alternatives

Chart	Feature
Treemap	Rectangular space-filling hierarchy
Circle packing	Nested circles for hierarchy
Flare chart	Radial node-link tree

Summary

The sunburst chart is a visually strong method for exploring hierarchical data. It is best used when the shape of the hierarchy matters more than exact value comparison.

References

Bump Chart

Tue, 30 Sep 2025 00:00:00 +0900

A bump chart visualizes changes in rank over time. Each line represents a category, such as a team, product, brand, or country. The horizontal axis shows time, and the vertical axis shows rank.

Unlike a line chart, which focuses on changes in value, a bump chart focuses on changes in relative position.

Historical Background

The idea predates the modern name. One early example appears in the 1890 Statistical Atlas of the United States.

This chart, edited by geographer Henry Gannett and others, connects the population ranks of U.S. states and territories across censuses from 1790 to 1890. It lets readers see not only rankings at each census, but also how each place moved over time.

Modern use of the term “bump chart” spread through data visualization practice in the 2010s, including Tableau examples such as Matt Chambers’s Car Color Evolution.

Data Structure

Category	Time	Rank
A	2020	1
B	2020	2
A	2021	2
B	2021	1

Rank 1 is usually placed at the top.

Use Cases

sports league standings
market-share rankings
school, company, or city rankings
election polls or popularity rankings

How to Read It

Each line is one entity.
The vertical position shows rank.
Crossings show rank reversals.
A line moving upward means the entity gained rank.

Design Notes

Put rank 1 at the top.
Limit the number of lines when possible.
Label lines directly at the right edge.
Use subtle grid lines for important ranks.

Alternatives

Slope chart for two time points
Line chart for actual values rather than ranks
Dot plot for static rank comparison

Summary

Bump charts are useful when the story is not the absolute value, but how relative position changes. They are especially effective for competition, ranking, and trend narratives.

References

Connected Scatterplot

Tue, 30 Sep 2025 00:00:00 +0900

A connected scatterplot connects the points of a scatterplot in sequence, usually over time. A normal scatterplot shows the relationship between two variables as a cloud of points; a connected scatterplot emphasizes the path through those points.

It is especially useful for showing how the relationship between two variables changes over time.

Background

The connected scatterplot builds on the scatterplot tradition and became especially visible in journalism and data storytelling after the 2010s. Media organizations such as The New York Times and The Economist used it to show the changing relationship between economic and social indicators.

Haroz, Kosara, and Franconeri describe the form as a way to present paired time series.

Driving Safety, in Fits and Starts — NYTimes.com

Data Structure

Variable	Type	Description
x	numeric	Horizontal variable
y	numeric	Vertical variable
t	time or order	Sequence variable

Points are ordered by t and connected with lines.

Purpose

The chart shows direction, cycles, loops, and changing relationships rather than static correlation alone. It is often used for economic, environmental, health, and social indicators.

Strengths

Shows both relationship and movement.
Creates a narrative path through data.
Can reveal loops or lagged relationships.
Works well when annotated with dates or arrows.

How to Read It

Each point is one time period. The line shows how the data moved from one period to the next. A line moving up and right means both variables increased; a line moving down and right means the x variable increased while the y variable decreased.

Design Notes

Use arrows or year labels to make direction clear.
Avoid too many series in one chart.
Keep axes and units clear.
Consider smoothing only when it does not distort the story.

Alternatives

Scatterplot for static correlation
Line chart for one variable over time
Animated scatterplot for time-based motion
Vector plot when direction should be emphasized

Summary

Connected scatterplots turn a two-variable relationship into a path. They are powerful for storytelling, but need careful labeling because loops and crossings can be difficult to read.

References

Polar Area Chart

Tue, 30 Sep 2025 00:00:00 +0900

A polar area chart represents values by the area of radial sectors. Each sector has the same central angle, while the radius or area changes according to the value.

The best-known historical example is Florence Nightingale’s coxcomb chart, which visualized causes of mortality among soldiers in the 1850s.

Historical Background

In 1858, Florence Nightingale created A Diagram of the Causes of Mortality in the Army in the East. She used visual statistics to communicate military mortality data to the public and policymakers. The chart is often called the Nightingale Rose Diagram and is considered an important origin of the polar area chart.

Today, polar area charts can be made with libraries such as Matplotlib and D3.js.

Data Structure

Element	Meaning
Angle axis	Category or time segment
Value	Radius or area
Category	Sector division

The format is well suited to cyclical data such as months, hours, seasons, or directions.

Use Cases

monthly sales, temperature, or rainfall
causes of death or accidents
24-hour traffic or energy use
seasonal disease patterns
weather and natural phenomena

How to Read It

Values grow outward from the center. Compare the size of each sector to understand relative magnitude. For area-proportional charts, the radius should be scaled by the square root of the value so that area, not radius, encodes the number.

Strengths and Caveats

Strengths:

Shows cyclic structure intuitively.
Has strong visual impact.
Works well with seasonal or periodic categories.

Caveats:

Precise comparison is harder than in a bar chart.
Too many sectors make labels crowded.
Small differences can be hard to judge.

Alternatives

Chart	Difference
Pie chart	Uses angle for part-to-whole proportion
Bar chart	Better for precise comparison
Radar chart	Connects values across radial axes

Summary

Polar area charts are useful when data has a circular or periodic structure. Their historical association with Nightingale also shows how visual design can support public argument and reform.

References

Wikipedia: Polar area diagram

Mosaic Plot

Mon, 29 Sep 2025 00:00:00 +0900

A mosaic plot visualizes relationships between categorical variables by representing cells as rectangles whose areas correspond to frequency or probability. The method was proposed by Hartigan and Kleiner in 1981 and later developed historically and theoretically by Michael Friendly.

The basic principle is:

Split the whole horizontally according to the first categorical variable.
Split each section vertically according to the next variable.
The resulting rectangle area represents the frequency or probability of that category combination.

By recursively splitting the space, mosaic plots can visualize more than two categorical variables.

Strengths

Cross-tabulation made visual: category combinations become visible as areas.
Conditional proportions: a plot can show both total group size and within-group shares.
Model diagnostics: colors can encode residuals from expected values in a log-linear model.
Exploratory analysis: variable order and category order can reveal or hide structure.
Education: the chart helps explain the relationship between counts and proportions.

Use Cases

survey analysis
marketing segmentation
statistical education
exploratory data analysis
model checking for contingency tables

Mosaic Plot and Marimekko Chart

A Marimekko chart or Mekko chart is structurally similar. It can be understood as a variable-width stacked bar chart. In business dashboards, the Marimekko name is common, while mosaic plot is the statistical term.

Perspective	Mosaic plot	Marimekko / Mekko chart
Context	Statistics	Business intelligence
Structure	Recursive area split	Variable-width stacked bars
Purpose	Cross-tabulation and model diagnostics	Market share and segment composition
Color	Often residuals or analytical meaning	Often presentation categories
Tools	R, JMP, S-PLUS	Tableau, PowerPoint, BI tools

Summary

Mosaic plots are a foundational statistical visualization for categorical data. Marimekko charts use a similar geometry in a business context. In both cases, the key idea is that rectangle area represents the size of a category combination.

References

Beeswarm Plot

Sun, 28 Sep 2025 00:00:00 +0900

A beeswarm plot shows a distribution by placing individual data points so that they do not overlap. It sits between a scatterplot and summary charts such as box plots or violin plots, preserving individual observations while revealing density.

The name comes from the way the points cluster like a swarm of bees.

Background

Beeswarm plots became familiar through the R beeswarm package and are now available in many visualization libraries, including Python’s seaborn.swarmplot() and Plotly strip charts. They reflect a broader trend toward showing raw observations rather than only statistical summaries.

Data Structure

Variable	Meaning	Type
category	Group to compare	string
value	Numeric measurement	number

The plot can show one distribution or compare several groups.

Purpose

The goal is to show the detailed distribution of data while retaining individual points. This is especially useful for sample sizes from dozens to hundreds, where every observation can still be displayed.

Use Cases

experimental or medical measurements by group
educational or survey score distributions
supplementing a box plot with raw data
SHAP beeswarm plots for machine-learning interpretation

How to Read It

One axis shows the numeric value, while the other axis separates categories. Points are moved sideways only to avoid overlap. The wider the swarm at a value range, the more observations are concentrated there.

Design Notes

Use transparency or sampling for larger datasets.
Adjust point size and spacing carefully.
Limit the number of categories.
Remember that horizontal spread is not a precise quantitative scale.

Alternatives

Chart	Use case
Box plot	Large samples and summary statistics
Violin plot	Smoothed distribution shape
Strip plot	Simpler point display for small data

Summary

Beeswarm plots balance individuality and distribution. Used with an appropriate sample size, they reveal outliers, clusters, and spread more directly than summary-only charts.

References

Parallel Sets

Sun, 28 Sep 2025 00:00:00 +0900

Parallel sets visualize multidimensional categorical data. Several categorical variables are arranged as parallel axes, and bands connect categories across adjacent axes. The band width represents frequency or proportion.

The method extends the idea of parallel coordinates from continuous variables to categorical combinations.

Background

Parallel sets were developed in the 2000s by visualization researchers Robert Kosara and Jorg Hauser. Their paper “Parallel Sets: Interactive Exploration and Visual Analysis of Categorical Data” introduced the method as a way to explore contingency-table data across many dimensions.

The method supports interaction such as reordering axes, rearranging categories, and highlighting subsets.

Data Structure

Element	Meaning
Axis	Categorical variable
Block	Category on an axis
Band	Combination between adjacent variables
Width	Frequency or proportion

Purpose

Parallel sets are used to understand relationships among several categorical variables, such as gender, age group, education, occupation, and income. They make multivariate cross-tabulations easier to inspect visually.

How to Read It

The vertical axes represent variables. Blocks on each axis are categories. Bands connect category combinations, and wider bands indicate more observations.

A key feature is that band width can be read as both:

a share of the whole dataset
a conditional share within a category

Design Notes

Axis order strongly affects readability.
Reduce or group categories when there are too many.
Use consistent color logic.
Interaction is valuable for filtering and highlighting.

Alternatives

Method	Data type	Feature
Parallel coordinates	Continuous variables	Numeric correlation patterns
Sankey diagram	Flows or categories	Directional flow
Mosaic plot	Categorical variables	Area-based cross-tabulation
Treemap	Hierarchy	Nested category area

Difference from Sankey Diagrams

Sankey diagrams usually represent directional flows such as energy, money, or process movement. Parallel sets represent combinations of categories. They are not necessarily directional, and relationships can be read from either side.

Summary

Parallel sets are powerful for exploring categorical data across many variables. Their clarity depends on careful axis ordering, category reduction, color design, and often interaction.

References

Radar Charts and Parallel Coordinates: A Debate About Readability

Sun, 28 Sep 2025 00:00:00 +0900

A radar chart, also called a spider chart, compares multiple variables by plotting values on axes radiating from a center point. Adjacent points are connected to form a polygon.

Radar charts are often used for student scores, product comparison, ability evaluation, and other profiles where viewers are expected to read balance or imbalance as a shape.

Weaknesses of Radar Charts

Radar charts are popular, but they are often criticized.

Radial comparison is hard: people are better at comparing lengths on a shared horizontal or vertical baseline.
Shape and area bias: the polygon can exaggerate differences that are numerically small.
Axis order is arbitrary: changing the order of variables changes the shape.
Limited scalability: many variables or many series quickly become unreadable.

Michael Correll’s essay “Bad Ideas in Visualization” argues that radar charts invite viewers to read shape even though that shape depends heavily on arbitrary axis ordering.

Parallel Coordinates as an Alternative

A parallel coordinates plot arranges variables as parallel vertical axes and connects each observation with a line. Compared with radar charts, all variables share a more consistent visual frame, which can make comparison easier.

The basic argument is that radar charts scatter axes around a circle, while parallel coordinates align them in a common direction.

Research Context

Research such as P-Lite: A study of parallel coordinate plot literacy compares how people read radar charts, parallel coordinates, scatterplots, and related views. The conclusion is not that one chart is always better. Rather, parallel coordinates can be more accurate for high-dimensional data in some tasks, while they also require more familiarity.

Popularity vs. Expertise

Despite criticism, radar charts remain far more widely recognized than parallel coordinates.

1. The Power of Shape

Radar charts turn multiple values into a memorable polygon. In presentations and education, that shape can be persuasive and intuitive.

2. Cultural Familiarity

Sports magazines, games, and school materials have made radar charts familiar as “ability charts.”

3. Tool Availability

Excel and Google Sheets include radar charts. Parallel coordinates usually require R, Python, or specialized visualization tools.

4. Learning Cost

Parallel coordinates can look tangled to non-specialists. Radar charts are easier to recognize at first glance.

Choosing Between Them

Radar charts can work when there are few variables and few items, and the goal is to show a rough balance profile. Parallel coordinates are better when there are more variables, more observations, and a need for more precise pattern exploration.

Summary

The question is not which chart is universally correct. The choice depends on audience, task, and data size. Radar charts communicate shape and balance; parallel coordinates support more systematic comparison and exploration.

References

The Origin of the Japanese Term "Quantity-Proportion Graph"

Sun, 28 Sep 2025 00:00:00 +0900

A quantity-proportion graph uses rectangular area to represent both absolute quantity and proportion. One dimension represents the size of a category, while the other dimension divides that category by composition. Each small rectangle’s area corresponds to quantity multiplied by proportion.

Internationally, similar methods are known as mosaic plots in statistics and Marimekko charts in business. The Japanese term “量率グラフ” emerged through education and popularization in Japan.

Examples

Total Amount Shown by Bar and Circle Area

The Same Data as a Quantity-Proportion Graph

Timeline of the Term

Mid-1950s onward: Kiyonobu Itakura later recalled using this kind of graph for about 30 years.
Late 1970s to early 1980s: Itakura introduced the graph in a study group on self-sufficiency rates. Hirotsugu Shiono proposed the name “quantity-proportion graph.”
Around 1982-1983: Shigehiro Matsuzaki became interested in the method and introduced it in elementary-school lessons.
June 1984: Matsuzaki published an article in Tanoshii Jugyo, where the term appeared clearly in print.
January 1985: Matsuzaki’s book Shakai o Minaosu Megane was published with the term in the subtitle.

Roles of the Three Figures

Kiyonobu Itakura: long-time practitioner who introduced the method in study groups.
Hirotsugu Shiono: proposed the Japanese name.
Shigehiro Matsuzaki: brought the method into classrooms and spread it through articles and books.

Mechanism

Vertical dimension: proportional to total amount
Horizontal dimension: proportional composition within the category
Rectangle area: amount represented by the intersection cell

This allows readers to see both volume and composition at once.

Naming

Mosaic plot: statistical name
Marimekko / Mekko chart: business-dashboard name
Quantity-proportion graph: Japanese educational name

The Japanese explanation often says “vertical = quantity, horizontal = proportion.” More precisely, both dimensions are relative scales, and the area is what encodes the resulting quantity.

Summary

The term “quantity-proportion graph” arose from Japanese educational practice in the 1980s. Although the chart can be understood internationally as a kind of mosaic or Marimekko chart, its Japanese history is distinctive because it spread through classroom use and educational writing.