Tips on Colour

by Charlie Daffern

Towards the end of last week DS7 were joined by Luke Stoughton who took us through a course on visual analytics. Colour is a huge part of any visualization, and should always be an active choice. Before I blow your minds with some great tips*, here are definitions from google for HUE and INTENSITY – two key pieces of language for thinking about colour.

HUE: ‘The attribute of a colour by virtue of which it is discernible as red, green, etc., and which is dependent on its dominant wavelength and independent of intensity or lightness.’

INTENSITY: ‘The measurable amount of a property, such as force, brightness, or a magnetic field.’

Varying hue and intensity are two ways that we as visualisers of data can really usefully employ colour in our visualisations. Discrete variables can be clearly distinguished using different hues, whilst intensity is a great way to illustrate variation in the concentration or volume of data represented by a single mark. Colour can be used alongside other features such as mark size to encode multiple measures within each mark. It’s great.

Here are a few specific use-cases where your choice of colour might be particularly important to your ability to convey a clear message.

Continuous variables not spanning zero
TO USE: Sequential colour scale.
WHY? Viewers can associate similarly coloured marks. If you use varying intensities of a single hue, this can be a perfect analogue for changes in the quantity of the measure represented across marks.

Continuous variables spanning zero
TO USE: Diverging colour scale, with the centre point fixed on zero.
WHY? Viewers are easily able to identify marks which are above and below zero based on the hue of the marks.

Discrete variables
TO USE: Non-sequential, but ideally also non-clashing, colours.
WHY? Using sequential colours for unrelated marks provokes an association response in the viewer, which might lead to erroneous conclusions about the way the data is presented. This example works because non-neighbour colours from the hue circle have been used – it wouldn’t work if the six variables had each been assigned a colour from the first column.

Beyond the theory, colour usage can be a very personal choice. Ask a room full of people what their favourite colours are and you’ll get a variety of responses. Even if you get a few people replying ‘blue’, if you ask them to pick the best colour from a series of blues you’ll probably get yet more divergence in the responses. Personally, I’m not a fan of pastel colours, and tend to try to use bold colours in my work – but of course, this isn’t the case for everybody. Targeting colours for your intended or expected audience is important. Branding of corporate material is a prime example of this. If you’re doing work on Nestle chocolate, you probably want to avoid using any rich shades of purple…

Now, people are probably not mega keen to see another viz in Tableau’s automatic blue palette – so what are a couple of top picks for colour scales?

Two that stand out are Viridis (pictured bottom) and Magma (pictured second), both of which David recently introduced me to. They are both beautiful and useful. Gwilym wrote a useful blog about Viridis, including some of the basis for why it’s so great, as well as the hex codes for adding it to your Tableau colour palette. David also recently shared with us the Magma colour palette code, which I’ve included here in case anybody can make use of it. Enjoy!

<color-palette name=”Magma 20 diverging” type=”ordered-diverging” >


*mind-blowing not guaranteed