Skip to main content
It looks like you're using Internet Explorer 11 or older. This website works best with modern browsers such as the latest versions of Chrome, Firefox, Safari, and Edge. If you continue with this browser, you may see unexpected results.

Data Visualization

Tips, tricks and tools for visualizing data

Visual Encoding

When designing a visualization, we encode into visual elements called marks and channels.

Marks

“individual items, or links between them” (Munzner, 2015, p. 95); points, lines, areas, containment or connections

Examples of marks: points, lines, areas, containment and connection

Image credit: Munzner, 2015, p.100

Channels

Represent attributes or dimensions of each datum; position, color, shape, tilt, size

Examples of channels, such as position, length, color and shape

Image credit: Munzner, 2015, p. 96

 

Expressiveness and Effectiveness

We choose marks and channels based on their expressiveness and effectiveness.

Expressiveness

“Visual encoding should express all of, and only, the information in the dataset attributes” (Munzner, 2015, p. 100)

Effectiveness

“…the importance of the attribute should match the salience of the channel; that is, its noticeability” (Munzner, 2015, p.101)

Expressiveness and effectiveness vary by dataset. They are also affected by the strengths and limitations of human vision. The following channels are ranked by most effective to least effective, depending on whether the data represents magnitude (“how much”) or identity (“what or when”). For a more nuanced ranking, see Mackinlay, 1986.

Recommended channels by data type

Image credit: Munzner, 2015,  p. 102

Data-Ink Ratio

Tufte (1983) defined data-ink as “the non-erasable core of a graphic, the non-redundant ink arranged in response to variation in the numbers represented” (p. 93). Using data-ink, one can derive the data-ink ratio:

Data-ink ratio calculation

Image credit: https://infovis-wiki.net/wiki/Data-Ink_Ratio​

William Playfair’s early charts have a low data-ink ratio. For example, “Chart of Imports and Exports of England to and from all North America" expends a lot of ink on details that don’t represent data:

Chart of Imports and Exports of England to and from all North America by William Playfair

Image credit: Wikimedia Commons

Later Playfair charts have a high data-ink ratio, in that almost all visual elements represent data or aid readers in understanding the data. “Exports and imports to and from Denmark and Norway from 1700 to 1780 “ uses less ink to show information comparable to "Chart of Imports and Exports of England to and from all North America."

Exports and imports to and from Denmark and Norway from 1700 to 1780 by William Playfair

Image credit: Wikimedia Commons

Tufte, a minimalist, advocated:

  • ”[Maximizing] the data-ink ratio, within reason” (p. 96)
  • “[Erasing] non-data-ink, within reason” (p. 96)
  • “[Erasing] redundant data-ink, within reason” (p. 100)

Deciding what to erase, or how to maximize the data-ink ratio, can be tricky because non-data-ink sometimes helps us to understand visualizations. Consider Tufte’s redrawing of “Cross-Enhancement of the Sour Taste on Single Human Taste Papillae" by James Kuznicki and N. Bruce McCutcheon (A). Tufte redraws the chart in B, eliminating non-data-ink and redundant-data-ink (C). The ink in accounts for 65% of the ink used in the original chart (A). Tufte argues that his version is just as effective as the original, but this remains open to debate.

 

Tufte redraws “Cross-Enhancement of the Sour Taste on Single Human Taste Papillae" to maximize the data-ink ratio

Image credit: https://goo.gl/images/ZP9swf​

References

Munzner, T. (2015). Visualization Analysis and Design. New York: A K Peters/CRC Press.

Tufte, E. (1983). The Visual Display of Quantitative Information. Cheshire, CT: Graphics Press.