A visualization should include as much information as possible, but not too much. Its visual style should be appropriate for the context, and it should be carefully executed down to the smallest detail. At its best, a visualization gives an overview of the topic quickly, but also rewards the reader who spends time exploring it in depth. - Koponen and Hildén
Data Visualization Handbook (2019), by Juuso Koponen and Jonatan Hildén, two Finnish academic authors, is a refreshing and welcome addition to the data visualization literature. Written as a handbook, it is a concise summary of the state-of-the-art in data visualization. More importantly, it includes a broad definition of data visualization and an expansive European perspective to a literature that is dominated by Anglo-American views.
The authors have carefully crafted the book as a handbook that can be easily digested and referenced. It is broadly divided into three parts: first, an introduction to vision, visual perception, and visualization design; second, an exploration of different information design genres, including information illustration, maps, statistical graphics, concept graphics, network diagrams, scientific visualizations, and text; and third, a focus on the information design workflow and ethics. The descriptions are detailed, concise, and clear. Each topic is illustrated with full-color images. And the authors point readers to the state-of-the-art research and researchers.
The first three chapters, Introduction, Visual Perception in Action, and General Principles of Visual Design provide an excellent foundation for data visualization and alone are worth the purchase of the book. They summarize the state-of-the-art and provide the background necessary to fully understand the issues in data visualization.
The authors start by establishing the importance of visualization, noting that research by Manfred Zimmerman shows that “our visual system sends our brains around eight times more information than all the other senses combined.” They identify the grammar of information graphics before describing the three defining actions that determine how easy or hard a visualization is to read: simplify, compare, and organize. They use this foundation to define the golden rule of information design: “choose the clearest presentation method available.”
The authors then describe the visualization process using Colin Ware’s three-stage model of the human visualization system. They describe the tunable or pre-attentive features of shape, color, and motion; showing how these can help readers spot elements, detect boundaries, follow moving elements, and estimate quantities. They move on to discuss the seven most widely accepted Gestalt laws, which explain how features are formed in the mind from their pre-attentive components. An example of this is the Law of Similarity, which states that “Elements that are similar to each other, for example with regard to color, size, or shape, tend to be perceived as belonging together.” This is an important topic that often receives short shrift in other data visualization books. Building on this foundation, the authors describe the visual variables, which are the building blocks of graphic visualization. This section concludes with a discussion of the many complexities of color.
At this point, the reader is ready to learn about the general principles of visualization design. Consistency is critical to a visualization where comparison is a focus. Consistency applies to the selection of colors, scales, symbols, and organization of the graphic. Layout refers to the arrangement of elements. The authors observed that “Layout is one of the basic building blocks of graphic design – the others being typography and the use of color.” Layout applies to simple graphics, as well as what the authors term “combination charts,” which combine statistical graphics, text, maps, or other illustrations. This section concludes with a discussion of the organization and categorization of data, including types of variables and scales.
The next major part of the book, “Information Design Genres,” takes an expansive look at types of graphics that go well beyond the typical description of statistical graphics. The authors not only describe the different types of graphics, but they also impart often overlooked conventional wisdom about the best practices for generating the graphics. This part is especially helpful, as it will give the reader insight into the wide range of possibilities for data visualization.
This discussion starts with a section on “Information Illustration” that covers a variety of solutions not typically found in data visualization books, including descriptions of annotated photographs, artist’s renderings, field guide illustrations, diagrams, and pictograms. The section on pictograms is especially detailed, discussing the design of pictograms as well as the impact of cultural stereotypes in their design.
This “Information Illustration” discussion moves on to technical drawing, where examples of axonometric projections (Isometric drawing, Dimetric drawing, and Trimetic drawing) and oblique projections (Military projection and Cavalier projection) are illustrated and discussed. They finish the discussion by covering cutaway drawings and step-by-step diagrams.
The section on “Maps” provides a comprehensive, yet concise description of maps for visualizing spatial data. The authors introduce a wide variety of map types, including thematic maps, which are divided into flow maps, dot maps, and areal unit maps. The areal unit maps category is further divided into maps showing enumeration units, like countries, states, and counties, and grid maps.
This is followed by a discussion of different types of anamorphic maps or cartograms, which are used to normalize data. This is a common technique for presidential election maps, where states with the same number of electoral votes have the same area, instead of showing their geographic area. They then discuss map types like topological maps or map diagrams, such as the London Underground map, and axonometric maps, which produce a three-dimensional view of the built environment. Finally, they conclude with maps for wayshowing, including strip maps and satellite navigation devices, vicinity and visitor maps, and on-site or “you are here” maps.
After introducing the reader to different types of maps, the authors move on to map design. Topics here include basemaps, scale, generalization, orientation and north arrows, color, map symbols, and map projections. An entire book about map design is reduced to a few pages that highlight the key considerations in map design.
“Statistical graphics” summarizes topics covered by most of the North American data visualization books. All the basic chart forms are covered at the beginning of the discussion, including bar charts, dot plots, line charts, pie and donut charts, scatterplots, and pictorial unit charts. The authors go into detail on the many variants of these charts. For example, the line chart discussion includes references to stacked line charts, slopegraphs, streamgraphs, bump charts, cycle plots, sparklines, and parallel sets charts. For a more in-depth discussion of these topics, readers are referred to Jonathon Schwabish’s Better Data Visualizations, which was reviewed in March 2021.
The chart descriptions are followed by a discussion of chart design, which includes observations on the oft-overlooked issues of aspect ratio; ordering of elements; scales and grids; and labels, legends, and annotations. The reader is given specific advice, such as the rule called “banking to 45o” for line charts which states that “differences in lines’ slopes in a chart can be best detected when the average slope is 45 degrees.”
“Concept graphics” identify a “loosely defined group of visualizations primarily showing conceptual and qualitative information.” It starts with an interesting discussion of the periodic table, showing alternative representations like the left-step periodic table, the zig-zag-shaped system, and the spiral-shaped “periodic snail.” This reinforces that there are often many ways to display data, which may vary significantly depending on the designer’s goal. The authors then go on to review matrices (or tables); quad charts; Venn diagrams; timelines; word clouds/bubbles; network diagrams; tree structure diagrams; Sankey and alluvial diagrams, as well as Marimekko/mosaic plots.
“Scientific visualizations” concludes the discussion of graphic genres. The authors define these as graphics typically derived from three-dimensional measurements or observations but concede that these can overlap with previously discussed visualizations.
“Text and Typography” receive special treatment. The authors review the basics of typefaces and discuss the details of variants, weights, and widths. This is followed by an assessment of the effect of typeface on legibility and readability and a guide to choosing a typeface. They move on to text typography, which they use as a “catch-all term for all other typographic details except the typeface itself.” This includes topics such as line length, line spacing, letter and word spacing, alignment, text direction, and the proper use of alignment in tables.
Koponen and Jonatan Hildén conclude their book with a chapter on “Information Design Workflows.” They identify three roles for individuals in the design process, including the Content Owner, Information Designer, and Implementer. Depending on the project and organization these roles may be taken by a single individual or a group of teams.
After examining Ben Fry’s and Moritz Stefaner’s models for the data visualization process, they propose a new iterative model. The Content Owner will: Define, Find & collect, Explore & organize; Sketch & experiment. The Information Designer will then Produce & refine the visualization. And finally, the Implementer will Assess the visualization and then Update & expand it. This process is illustrated with a description of the New York Times graphics desk.
The authors give simple advice to designers
Stick to established chart types, unless a novel presentation method is superior
The data must be correct and processed correctly
All details in graphics should correspond to reality
They conclude with a series of ethical guidelines for visual journalists as well as for journalists in general.
The two appendices cover the topics of map projections and interaction. The topic of interaction reviews general design principles for interaction, basic interactions, and interactions typical of visualizations. The discussion about interaction will be especially helpful to the reader, as this topic is typically not covered in data visualization texts, which primarily deal with static graphics.
To conclude, this is a book that should be on every data scientist’s bookshelf. It is a comprehensive and concise summary of data visualization, covering topics ranging from vision and the grammar of graphics to visual genres and workflows for data visualization. Readers will be introduced to information genres that are not typically covered in data visualization texts, including information illustration, concept graphics, network diagrams, and scientific visualizations. The authors cite state-of-the-art research, whether it is from the more widely known North American literature or the less familiar European literature.
Data Visualization Handbook Website (contains tutorials and posters)