Definition:
Data visualization is the graphical representation of quantitative or qualitative information through elements such as position, length, size, color, shape, and connections. Its purpose is to reveal comparisons, changes, distributions, relationships, or structures that would be difficult to recognize in a table or a set of isolated values.
A visualization is not merely a decorative image. Each graphic element encodes a variable or relationship: the height of a bar may express a quantity; the position of a point, two values; and the color of an area, the intensity of a measure. Correct interpretation requires knowing the data source, units, scales, categories, and period represented.
The discipline predates computers and Web 2.0. William Playfair’s statistical charts, John Snow’s cholera map, Florence Nightingale’s diagrams, and Charles Joseph Minard’s chart of Napoleon’s campaign are historical predecessors of modern visualization. Software later expanded the capacity to process large datasets and create interactive charts, but it did not originate the concept.
Table of contents
What a visualization can show
The appropriate graphic form depends on the question and the type of data available. No single chart works equally well for every analysis.
- Comparisons: bar charts contrast values across categories. Length makes it easier to compare magnitudes when all bars share a common baseline.
- Change over time: line charts show how a variable changes over time. The frequency of the data and the selected interval affect the shape of the series.
- Distributions: histograms group observations into intervals; box plots summarize location, spread, and outliers. Both show how data are distributed rather than only reporting an average.
- Relationships: scatter plots position observations according to two variables and can reveal associations, clusters, or outliers. A visual association does not by itself establish causation.
- Composition: stacked charts and other part-to-whole displays show how a set is divided. The total and the grouping criteria should be explicit.
- Location: maps connect values with geographic areas, points, or routes. Region size and data normalization can affect perception.
- Hierarchies and networks: treemaps, trees, and node diagrams represent levels, links, or flows between elements.
An infographic may combine charts, illustrations, and text within a visual narrative, but not every infographic contains data and not every visualization is an infographic. A heat map, meanwhile, encodes intensity through color and can represent anything from values in a matrix to activity on an interface.
From data to chart
Before a chart is drawn, someone decides which observations to include, how to group them, and which transformations to apply. A sum, mean, percentage, or rate answers a different question. Missing data, category comparability, and whether the dataset actually represents the phenomenon also matter.
Visualization translates these decisions into visual properties. Position and length usually support more precise comparisons than area or volume. Color can distinguish categories or display a scale, but it needs a legend and should not be the only cue when viewers must tell multiple series apart.
Axes and scales are part of the meaning. Truncating the axis of a bar chart can exaggerate differences; a logarithmic scale changes the visual distance between values; and combining two vertical axes can suggest relationships that depend on how those axes were adjusted. Titles, labels, sources, and notes should explain these decisions when they affect interpretation.
Descriptive analytics uses visualizations to summarize what observed data show. In a broader analytical process, a chart can also help explore hypotheses, detect errors, or communicate results, but it does not replace statistical validation or repair quality problems in source data.
Static and interactive visualizations and dashboards
A static visualization presents a defined view, such as a chart in a report, news story, or presentation. It can be carefully annotated and works without interaction, although it only shows the selection made by its creator.
An interactive visualization allows people to filter, sort, zoom, change variables, or inspect details. These actions support several perspectives on the same dataset, but active filters must remain visible and navigation must be understandable with a keyboard, mouse, or touchscreen.
A dashboard brings together several metrics and charts to monitor a specific process or area. It is not a chart type but a composite interface. In business intelligence, dashboards often connect to data sources and update indicators, dimensions, and filters according to system rules.
Tools range from spreadsheets and business intelligence platforms to programming libraries and statistical languages. Microsoft Excel, Power BI, Tableau, Looker Studio, R, Python, and D3.js can all produce visualizations, but they differ in data transformation, automation, interaction, publishing, and control over design.
How data visualization is evaluated
An effective visualization should be faithful to the data, legible, and appropriate for its context. Its quality does not depend on being visually striking, but on enabling interpretation without hiding uncertainty or encouraging misleading comparisons.
Checks should include the source and freshness of the data, the definition of each metric, units, percentage denominators, and filtering criteria. Missing values, estimates, margins of error, or methodological changes should also be identified when relevant.
Legibility requires sufficient contrast, clear text, and a visual hierarchy that does not compete with the data. Excessive colors, three-dimensional perspectives, or decoration can make comparisons harder. People with color vision deficiencies need additional cues such as labels, patterns, or shapes.
On the web, a complex chart needs a text alternative that communicates its essential information. The W3C guidance on complex images covers short and long descriptions, data tables, and other equivalents depending on complexity. Web accessibility also requires interactive controls to work with a keyboard and information not to depend solely on color or motion.
Data visualization turns values and categories into a visual system for comparison. Its interpretation depends both on the chosen graphic and on decisions made before drawing it: selection, aggregation, scale, context, and explanation of the data.