Skip to Content

This blog is part of a series to highlight useful features, tips and tricks to assist with good data visualization practice in Lumira 2.0 Discovery.

The first two blogs in the series can be found here:

SAP Lumira 2.0 Visualization Tips & Tricks: Adding Context with Reference Lines

SAP Lumira 2.0 – Visualization Tips & Tricks: Adding Context with Conditional Formatting

 

Less is More!

 

The great modernist architect Ludwig Mies van der Rohe, is often credited with coining (or at least popularizing) the phrase “Less is More”. Rarely is this axiom more appropriate than in the art of good data visualization.

We are increasingly inundated with huge volumes of data and with increasing capabilities with which to show such data – as such, it is easy to fall into “traps” to over-elaborate our data visualizations and cause the essential meaning of our data to be hidden or lost.

SAP Lumira allows great power, flexibility and features to craft your data visualization and add “embellishments” – but exercise due caution and remember often that Less is More!

To illustrate this, consider the following set of company product sales (“Units Sold”) data, which is shown in a simple Lumira crosstab visualization:

 

 

We could leave this data as a crosstab but it can look fairly “plain”, so the temptation is to represent it in another type of data visualization, for example a column chart.

With flexible self-service BI tools like SAP Lumira, it is very quick and easy to add data visualizations and to “enhance” these with many (apparently) cool and colorful features.

However, care should be taken as, without due consideration, the essential meaning of the chart data can become quickly obscured and difficult to extract.

Hopefully no one would produce anything as hideous as this particular column chart! 😉  – but it can be done in Lumira and serves to point out some easily acquired traps/pitfalls in visualization techniques.

 

 

So, some obvious criticisms of the above chart visualization include:

 

  • The multiple chart colors, especially the background color, are too dominant and overpower the chart data
  • The chart columns are too narrow and confusing
  • There are no specific values to any of the columns
  • The chart data is not in any specific order – should it be ?
  • The chart title is too vague
  • And so on…

 

Let’s see what can be done to improve the visualization – and how this is achieved in Lumira:

 

Remove the Background Color

 

It is tempting to use background color to “highlight” the chart but this can actually result in the chart data becoming more obscure

  • The blue background is too similar to some of the column colors and masks them – the “Sweatshirts” column is almost invisible against the background
  • The yellow (Coats) and orange (Vests) columns are prominent against the blue background – why? Is there some special significance to these columns?
    • Use the “Squint Test” – squint your eyes to see which chart elements stand out visually – are they the ones that are meaningful and appropriate?

 

To remove or adjust the background color, go to the “Design” tab on the left of the screen and navigate down to Set Visualization Properties –> Chart Area (dropdown) –> Background Color (dropdown)

(Repeat this also for the Chart Plot Area).

 

 

>>> Note, the use of color is a subjective and often controversial topic. If background colors are mandated, for instance due to corporate branding standards, consider using one of the many free resources that are available to help you choose appropriate color schemes – for example Adobe Color or Color Scheme Designer.

 

By removing the blue background, the chart content is immediately more clear:

 

 

However, the improvement in clarity is still not enough.

 

Remove the Chart Legend and Color Dependency

 

In this particular case, the chart legend is unnecessary as we are only looking at one dimension for the columns and having a legend forces the viewer to make the association between column and color

  • Which column represents Sweaters, which represents Sweatshirts – it is difficult at first glance to determine, they are both blue

 

To remove the legend – again, use the “Design” tab:

 

 

This further improves the chart – removing the legend/color-dependency also increases the column width as a by-product:

 

 

We’re getting there – but it’s still not quite enough…

 

Data Labels

 

The bold chart grid lines may seem to be a good mechanism to help indicate the chart values but It is difficult to determine the exact column values just by using the grid lines – it would be better, in this case, to explicitly show the chart column data values. So…

  • Add the data labels
  • Remove the grid lines

 

The data values can be added and the grid lines removed via the “Show/Hide Properties” menu, that is available by right-clicking on the visualization (“Show/Hide Properties” gives quick access to the most commonly used visualization properties).

 

 

This modifies the chart visualization to appear as follows:

 

 

The chart clarity has further improved…

And clearly now, displaying the column data values invalidates the need for the “Max Value” reference line – this can be removed by right-clicking on the chart again and selecting the “Reference Line” option.

 

 

The chart has now been simplified significantly but, at the same time, has become much more clear in its depiction of its data

We are nearly there – but there is still some further simplification and clarification that we can do…

 

Sort – and Bring Order!

 

This above visualization is a great improvement over the original – but the data is not in any particular sort order.

We can bring further clarity by sorting it – (to do this right-click on the visualization and then select (for example) a descending sort).

 

 

The result of sorting, implicitly shows the ranking of the products in terms of Units Sold.

 

 

This further invalidates the need for the previous “Max Value” reference line as the viewer’s eye is automatically drawn to the product with the maximum quantity of units sold (Shirts) as well as highlighting its relative rank in relation to the other products.

 

Finally – applying a more meaningful chart title brings additional clarity to the chart’s purpose (update the title by simply double-clicking on it and overwriting with the desired text).

 

 

While the above chart visualization is less colorful and thematic than the original, I hope the point is made that it is much better at conveying the key insights of its data in a rapid, and understandable fashion.

Removing the chart clutter, removing the junk is often key to improving the visualization’s impact and insightfulness. Less is (often) more!

 

 

Additional useful references for good visualization best practices in SAP Lumira can be found here:

Creating High Impact Data Visualizations with SAP Lumira” – a recording of the webinar, presented by SAP’s Jacob Stark (for Lumira 1.x).

The “SAP Lumira Data Storytelling Handbook”.

 

To report this post you need to login first.

7 Comments

You must be Logged on to comment or reply to a post.

  1. Moritz Kelm

    Good article, but one addition: When the value labels are activated and bars sorted, there’s no need anymore for the value axis on the left, right? 😉 This cleans it up even more.

    (1) 
    1. Adrian Paul Post author

      Thank you Moritz.

      Great point – you are correct in that the left axis values could also be removed since the column data values are displayed. However, I left it there as it provides a sense of “scale” to the displayed values – i.e. especially in relation to the top axis value of 15k.

      (1) 
    1. Adrian Paul Post author

      Hello Clemens,

       

      I’m not sure that I completely understand your question – but you can create the chart visualization/design and then point it to different datasets in order to display different data values.

       

      Best Regards,
      Adrian.

      (0) 
  2. Sebastian Ospina

    Hi Adrian.

    I am trying to change the label of my values ​​format for a bar chart, but it is not possible (Lumira Discovery), I did the same exercise for a table and if it works I can convert to thousands and remove the decimals, but in graph I I could not, you can help me. Thank you.

     

    (0) 
    1. Adrian Paul Post author

      Hello Sebastian,

       

      If you create the Crosstab (table) chart first and then convert it to a bar chart (using the Chart Picker right-click option) the Data Values should reflect the new format.

       

      Best Regards,

      Adrian.

      (0) 
  3. Saurabh Sharma

    Hello

     

    Please let me know what is the migration strategy for Migrating the already existing Lumira to new host. We have to take our old Lumira to old Host, we have to take it to new host.

    Thanks

    Saurabh

    (0) 

Leave a Reply