An official website of the European Union

An official EU website

European data

data.europa.eu

The official portal for European data

Visualisation Notes #1

The Power of Data Visualisation

How visualisation allows us to reason through data

Published on 25 November 2024

We hope that you've enjoyed the first data story of our series based on the European Union's (EU) high-value datasets!

In this story we used maps to show areas in Spain that may be at risk of fluvial or coastal flooding in the next years or decades.

Maps that represent data are part of a practice called data visualisation. Our goal with this series of stories isn't just to create beautiful graphics based on high-value datasets, but also to prove that visualisation is a powerful tool to reveal patterns and trends in data — and to teach you how to use it.

Therefore, each of our data stories will be paired with an article like this, intended to introduce you to the wonderful world of data visualisation.

Let's begin with some basics.


Hans Rosling, a popular Swedish international health professor, medical doctor, book author and statistician once wrote that 'The world cannot be understood without numbers. But the world cannot be understood with numbers alone'.

Quantification is an essential part of science and critical thinking, but it's not sufficient on its own. Qualitative knowledge and domain-specific expertise are necessary to put data into context.

Moreover, when it comes to communicating insights from data, there are few tools that are as useful as data visualisation.

Data visualisation consists of encoding data. This sounds a bit technical, doesn't it? However, it isn't hard to understand.

Here's an example: when you ask a software tool to create a bar graph like the one below, what is the tool actually doing?

A simple bar chart: 5 light blue bars aligned at the bottom and a numerical
      vertical axis on the left. The number ticks on the axis read, from bottom to top: 0, 5, 10, 15, 20, 25. The heights of the bars
      are equivalent to the numbers at the top of each bar. From left to right, they read: 12, 25, 18, 8, 3.
Figure 1: A bar graph with simulated data

The software tool first reads your data and connects it to a series of geometric objects.

In the case of the bar graph, these objects are rectangles:

The same bar chart from the previous image. Except, now all the bars are the same height and the
                            numbers are disconnected from them. The image is divided in three parts. The text at the top of the image reads: 'Data that we want to visualise: 12, 25, 18, 8, 3'.
                            At the middle section of the image reads, five arrows connect the numbers to the bars at the bottom, with text that reads: 'the tool connects them to...'.
                            At the bottom of the image, the bars are accompanied by text that reads: 'A series of objects (rectangles).
Figure 2: A bar graph where all bars have the same height because they haven't been varied based on the numbers associated with each of them

The software tool then changes the height of these rectangles in proportion to the numbers that we want to visualise.

The same image as the previous one. But now, the bars have their heights correspondent to the numbers 12, 25, 18, 8 and 3.
Figure 3: A bar graph where the height of each bar is now proportional to the quantity it represents

This process of connecting data to graphical objects is called encoding. Visualisation is a language and encoding is the foundation of the grammar of visualisation.

In the grammar of visualisation, the objects we use to represent data — dots, rectangles, circles, triangles, lines and others — are called marks.

A collection of geometrical shapes: squares, circles, semicircles, lines and dots.
                            The title of the image reads 'Examples of marks'.
Figure 4: Data visualisation can use many types of 'marks', geometric symbols.

The attributes of these marks that we vary to represent our data (e.g. the height of the bars in the case of a bar graph) are called visual channels.

Visualisation designers use a broad variety of visual channels to encode data. For example, we can vary height or length (like in bar graphs), angles (like in pie charts), areas (like in bubble charts), colour hue and shade (like in certain types of data maps) and many others.

The title of the image reads 'Examples of visual channels'. Below it, an abstract
        bar chart reading 'Height'. A collection of circles with varying diameters reading 'Area'. An abstract pie chart reading 'Angle'.
        An abstract scatter plot reading 'Position'. An abstract line chart reading 'Also position'. A trio of differently shaded squares
        reading 'Color'.
Figure 5: Visual channels in visualisation are the attributes of objects (marks) that we can vary to represent data.

Now that we've learned a bit about the grammar of visualisation, it's time to go back to an example from our story about flooding risk in Spain. This is one of the data maps in it:

One of the maps from the main story. It shows the area at flood risk in the Guadalquivir River basin, in Spain.
        Throughout this area are scattered a bunch of dots in three different shades of purple, from lighter to darker. A color legend reads
        'Points at risk: Light risk for light purple, grave risk for medium purple, and very grave risk for dark purple'.
Figure 6: A map that appears in Story 1 and shows points at risk of flooding near the Guadalquivir River basin in Spain

Let's try to identify the marks and visual channels used in this map.

One visual channel is position. A map is generated by positioning points, lines and other marks over a coordinate system, where the horizontal axis corresponds to longitude and the vertical axis corresponds to latitude.

This map also uses another visual channel - colour - to identify different types of geographic features, such as the points at different degrees of risk of flooding.

The power and purpose of data visualisation lie in helping us notice patterns that may be impossible to see if the data was presented just as figures in a table. Imagine that we only showed you the data itself, but not the map; you wouldn't understand much:

A screenshot from the data table associated with these data points. It contains dozens of columns
    with a bunch of different information, too much to make sense at a glance.
Figure 7: A screenshot of the dataset behind the map in Figure 6

But by transforming the data into a visualisation — a map, in this case — your eyes and brain can quickly perceive where the points at high risk of flooding are more or less concentrated. That's the power of data visualisation.


That's all for now! In following articles, we'll discuss other aspects of data visualisation. For example, how to choose the most appropriate type of chart for every message or story, or the fact that individual visualisations are sometimes meaningless on their own and acquire deeper meaning when woven into a narrative with other visualisations.