Chart to success: confusing colours

We live in a time when data is king. Data is everywhere – and it’s tempting to believe that, the more data we have, the more accurate our decisions will be. But regardless of the data, even the most simple graphs can easily mislead you. Even when using the best tools available, your conclusions may not be as sound as they seem if you can’t spot some fundamental issues. 

In this series, we’ll be taking bite-sized looks at some of the most common data analysis mistakes made when building and interpreting graphs. 

Even the colours of your graph can mislead you. That’s right – sometimes, you can’t even trust your own eyes!

While a rainbow heatmap can look cool, it isn’t perceptually uniform. That means your eyes interpret the rainbow colours with different “intensity”, making you see groups that simply don’t exist. Essentially, two values that should be appearing to be very close together actually appear to be very far apart. Further, by using the right colours, you can unveil more detail!

Consider these two graphs of the same datasets:

You can probably see some clear & distinct groupings in the left chart. These actually don’t exist! They’re much more subtle (and fairly represented) on the right, simply because of the different colours. Look specifically at the blue border on the left graph that seems to join the two blobs (around x=50 / y=50) – our graph on the right shows the previously hidden divide between the two. We’d have missed this if we hadn’t used a good colour scale!

While the exact colours used are subject to some fancy science which is too detailed to get into here, you can always defer to pre-programmed colour maps like Viridis or many others. Most good analytics tools will provide these as an option.

The added benefits to using these perceptually uniform colours include:

  • They usually appear similarly if you’re colourblind (depending on the type)
  • They usually print out the same in greyscale too.

Recommendation

Before you use or view a heatmap, consider if other graphs could tell the story better (you should always do this!).

If the data really is best suited to a heatmap, make sure you consider the colours used, and make ensure they’re well-designed and perceptually uniform.

If you’re using graphs to make data-driven decisions, it’s important that they are accurate and reliable. Graphs can be misleading if the graph type, colours or scales are poorly chosen, so look out for these details before making a decision.

There is no such thing as a “perfect” graph – every graph has its limitations! However, there are many ways in which we can improve our understanding of any given dataset by carefully considering all aspects of the analytical process.

Vaxa Analytics offers free analytics audits to help elevate your business intelligence with insights from experienced professionals who understand what makes a good analysis tick.

Contact us today for more information on how we can work together to create data-driven, actionable insights to help your business succeed.

Wouldn't you rather know? automate? grow? innovate? analyze? personalise? excel? accelerate? maximise?

Let’s turn the hidden potential in your data into a catalyst for your organisation’s transformation.

More thoughts

Image depicting a filing cabinet, with the word Vaxa spelled letter-by-letter on the labels of the files.
Insights
Curtis West

SMEs and data governance: an approachable guide

Many SMEs struggle with understanding and implementing effective data governance practices which underpin all data-driven decisions. In this approachable guide, we’ll take a look at what data governance is, why it’s important for SMEs and some best practices for getting started.

Read More »