The origins of data visualisations

These days, everyone is a bit of a data analyst. At primary school, you learn to read and fill in tables. At secondary school, creating line charts is often one of the first things you learn in maths. But you also encounter data visualisation every day outside school or work: bar charts and line charts appear in news reports almost daily, and many adverts use them to reinforce their message. A good visualisation helps you communicate information efficiently. A line chart, for example, shows at a glance what is happening with coronavirus infections in the Netherlands, without you having to compare figures or do any calculations yourself. If the line rises, infections are rising; if it falls, the infection rate is falling. This lets you use intuition and insight to draw a quick conclusion from a large amount of information. Almost anyone can read different types of visualisation quickly and accurately, without prior knowledge. But that has not always been the case. In fact, data visualisations as we know them only began to emerge a few decades ago. So where did it all begin? And where is it heading?

The first visualisations

The earliest known data visualisation is a wall painting in a Turkish cave. Estimated to date from 6200 BC, it is a map of the Neolithic village of Çatalhöyük. Alongside the village buildings, it shows a nearby volcano that appears to be about to erupt. A map might not be the first thing you think of when you hear ‘data visualisation’, but it is essentially a graph with an x-axis and a y-axis. On this graph, each building (seen from above) has a set of coordinates showing exactly where it is.

The cave drawing at Çatalhöyük is considered the world’s oldest map. Since then, maps have become increasingly precise and standardised. The most obvious reason is that measuring equipment has become much more accurate, but the way maps are laid out has also become largely consistent. Almost all maps now have a scale, a compass rose showing where north is, and a grid with corresponding coordinates or markings.

Figure 1. The cave drawing at Çatalhöyük1

William Playfair – the first line chart

Alongside maps, line charts were among the first forms of chart as we know them today. The earliest line charts date from the end of the Middle Ages and the 16th century, when they were used mainly to record the positions of stars. Yet William Playfair, an 18th-century Scottish engineer and political economist, is often regarded as the founder of statistical graphics. In 1786, he designed the line chart as we know it. He also invented the bar chart and the area chart, and later the pie chart.

Playfair’s line chart plots two variables, England’s imports and exports with Denmark and Norway, against time. It contains all the elements of a modern line chart, including labelled axes, a title and a brief explanation of what the axes represent. But some stylistic choices make the chart less clear and would probably be handled differently today. For example, the meaning of each line is written directly on it, in a way that is rather hard to read. A legend could help here. The conclusion that can be drawn from the chart is also highlighted within the chart itself, using coloured areas between the two lines. Today, a conclusion is often included in the accompanying text or the figure caption.

In the following century, John Snow showed that a conclusion does not always need to be stated directly in a visualisation: it can instead be drawn from it.

Figure 2. Playfair’s trade-balance time-series chart2

John Snow – cholera

In the 1850s, cholera swept through London’s Soho district. At the time, bacteria and germs were not yet understood, and the sudden outbreak and its cause were a mystery. John Snow, a British scientist who later became a founding figure of epidemiology, decided to map every case of cholera. He suspected that water contaminated by waste might be to blame.

He marked every case of cholera on a map of the district. A pattern emerged: most cases occurred on Broad Street, forming a clear cluster. A water pump stood at that exact spot, and many local residents drew their water from it. This supported John Snow’s suspicion, and he traced the cholera outbreak to the pump on Broad Street. Snow took a water sample and examined it, but did not find enough evidence to prove that the contaminated water was responsible. Even so, the map was convincing enough for the local authorities to take action and disable the pump. This marked the end of the cholera outbreak in Soho.

This example is often mentioned in the history of data visualisation because it is one of the first known cases in which a conclusion drawn purely from a visualisation led to action. Ultimately, that is what data visualisation is about: presenting data in a way that creates new insights or helps people understand them. In John Snow’s case, plotting every case on a map helped pinpoint the source of the infection.

Figure 3. John Snow’s London Cholera Map3

The rise and future of data visualisation

Data visualisation has developed rapidly since the 1950s. The introduction of computers, in particular, has made many new forms of visualisation possible. It suddenly became possible to process larger volumes of data and predictions from computer models, leading to new forms of visualisation such as word clouds. As mentioned earlier, more traditional forms such as line charts and bar charts are still used daily in the news and in newspapers, partly because almost everyone can read them without much difficulty.

Today, data visualisation goes beyond simply displaying large amounts of data. Infographics, often a distinctive combination of several data visualisations, can explain a complex problem or topic by bringing together images, data visualisations and just a small amount of text. Another interesting development is that these infographics are often designed by visual designers, rather than solely by scientists as they were in the past.

When creating data visualisations, it is important to remember that readers see a visualisation as it was designed. To some extent, you therefore rely on the designer’s interpretation of the data, whether that designer is a data scientist or a visual designer. The more complex the data, or the more data brought together in a visualisation, the more easily misunderstandings can arise for both its creator and its readers. That is why you should always check for yourself whether a correlation really indicates causation, especially since a good visualisation is not necessarily based on a good analysis.

Personally, I think there is still much to gain in the future from bringing scientists and visual designers together, so that science can be communicated even more effectively. The first steps have already been taken: many companies are setting up data teams in which data scientists work closely with data analysts. The next step is to involve creative professionals as well.

References

1. http://www.sci-news.com/archaeology/science-catalhoyuk-map-mural-volcanic-eruption-01681.html
2. http://scihi.org/william-playfair-and-the-beginnings-of-infographics/
3. https://www.nationalgeographic.org/activity/mapping-london-epidemic/