Data Visualisation: A Handbook for Data Driven Design

by Andy Kirk

A waffle chart shows how proportions of quantities for different constituent categories make up a whole.

Stacked bar charts often work best when the categories are ordinal in nature, and it is the overall pattern of spread across the whole that is important.

Ideally a scatter plot will have a squared aspect ratio (equally tall as it is wide) to help patterns surface more evidently.

If one quantitative variable (e.g. weight) is likely to be affected by the other variable (e.g. height), it is general practice to place the former on the y-axis and the latter on the x-axis.

The geometric accuracy of the shape mark size calculation is paramount: it is the area you are modifying, not the diameter/radius.

line chart shows how quantitative values have changed over time for different categorical items.

A bump chart shows how quantitative values have changed over time for different categorical items, where the quantitative values are ranking measurements.

A slope graph shows how quantitative values have changed over two points in time for different category items.

A prism map displays quantitative values for locations on a map. The values are represented via proportionally sized lines, appearing as 3D bars, that typically cover a fixed surface area of space and are then sized through height to proportionally represent the quantitative value at each location.

Technological: What charts you can actually make and how easily you can personally create them is a big factor.

To assist with this, you might consider consulting the ‘Chartmaker Directory’ (chartmaker.visualisingdata.com/).

The profile in the (near) future will be a hybrid one, mixing competences, skills and approaches currently separated into disciplinary silos.’

Purpose: Having the technical ability to create a broad repertoire of chart types is the vocabulary of this discipline; judging when to use them is the literacy.

Do not rule out the value of a table if providing a means for your viewer to look up and reference values.

In his book Semiology Graphique, published in 1967, Jacques Bertin proposed the idea that different ways of encoding data might offer varying degrees of accuracy in the perception of data values.

Two classic illustrations of this notion are shown below. Looking at Figure 6.55, if A is 10, how big is B?

This is important to acknowledge because you have to weigh up whether precise perceiving is actually what you wish to offer your viewers.

I like the idea of making people say “oh that’s beautiful! I want to know what this is about!”

Exploratory data analysis, in many ways, offers this bridge to visual communication: the charts you use to see data for yourself often represent prototype thinking about how you might communicate real data to others.

Always give yourself time to spend on the editorial stage, carefully articulating what you want to say before you get too carried away with picking how.

Viewers base estimates of quantitative size through the area of a circle, not its diameter.

For charts genuinely based on three dimensions of data, a 3D representation should only be considered reasonable if the viewer is provided with the means to adjust the field of view.

Features of interactivity must be fundamentally justifiable. They must enhance and not obstruct the facilitation of understanding.

Do not pass on to the user the task of discovering insights if a context necessitates the provision of an explanatory experience. This often betrays a certain lack of commitment to editorial clarity.

As we encounter each option, it will be useful for you to understand the distinctions between an event, the control and the function:

This action effectively modifies the editorial ‘framing’ perspective for the current view of data.

In contrast to the filtering options, these functions modify the editorial ‘focus’ perspective.

Linking and brushing are common approaches used in exploratory data analysis tools where you might have several chart panels and wish to see how selected items compare across each display.

Asking people ‘what do you think?’ and providing immediate feedback to their response is a compelling way to challenge or reinforce people’s perceived understanding about a subject.

Passenger numbr quiz in the mornings. Applicable to any absolute metric.

Each question is framed the same way, asking how much of a given item can be consumed until six teaspoons of sugar have been reached.

Recognising that users may not necessarily understand how to read them, Bloomberg’s visual data team provide a pop-up ‘How to read this graphic’ guide when they visit the project shown in Figure 7.17.

Data that has a temporal dimension can present opportunities for being displayed using some form of animated sequencing.

One of the main benefits of interactivity is to overcome the limitations of space.

An alternative approach to navigating involves the offer of a more linear, explanatory experience, building up a narrative about a subject through a series of discrete sequences.

It is arguably the quintessential example of storytelling with data and is often presented using a technique known as ‘scrollytelling’, whereby users scroll to move vertically up and down through the steps of a story.

The project featured in Figure 7.22 is a prime exhibit of this kind of dynamic interface.

Placing buttons (filters) to he left

Sequentially building up a story can prove to be a powerful way of facilitating understanding.

As you scroll down the page (or, alternatively, click through the page steppers or directional arrow) it takes you through the different hypotheses, overlaying data about each onto a chart plotting the observed changes in temperature.

the sequence here gradually builds up the user’s understanding about the subject matter.

These offer ways of choosing a country to navigate to a separate report that provides far more detailed analysis and commentary, supplementing the summarising data as displayed in the initial flower view.

Dashboards that use CoE architecture categorisation.

Just because a visualisation might be created and published digitally, the output may still be non-interactive: digital does not necessarily or automatically mean interactive.

Will your audience have the time, the patience and indeed the know-how to exploit such features?

For instance, you might open a visualisation with an explanatory experience, based around showing some main findings and telling your audience something. Through interactivity, you may then transition the users towards more of an exploratory interface that invites them to interrogate the data to pursue their own particular curiosities about the subject.

Indeed, resort to interactivity only when you have exhausted the possibilities of an appropriate and effective static solution.

For a one-off project you have to rely on your own best judgement; for projects that will be repeatedly used you will have more potential to seek and accommodate feedback to inform refinements.

This concerns judging the level of assistance an audience may require in order to understand the background, function and purpose of a project,

judgements about what annotated features to offer your viewers can be more heavily informed by common sense.

The primary aim of a heading is to inform your viewers efficiently about the content they are about to encounter and to orientate themselves within the hierarchy of this content.

The statement title (Figure 8.1) might be most commonly used with visualisations that offer an explanatory experience, based on the mantra ‘if you have something to say, say it’.

A title presented as a question (Figure 8.2) can offer a compelling way to align your audience’s minds with the essence of the curiosity that has driven the project.