Explore-Explain

Is your goal to explore or explain your data or ideas, or a combination of these?


A common view on the essential reasons to visualize something is that we either want to explain it, or explore it. While these are not mutually exclusive, design choices might be very different if the main goal is to explain or to explore.

We use the explain category for mainly static, single visualizations that are designed for a particular purpose and audience (e.g., textbook illustrations, plots and maps for scientific publications, visualizations for printed media). This means they are typically summarized for the content and somewhat customized for the viewers. When creating an explanatory visualization, it is important that we show the most relevant information, while in the exploratory visualizations, what is relevant is left to the viewer.

Example for ‘explain’: Can you follow the instructions?
Below is a typical visualization that explains how to do something, the familiar furniture assembly instructions (source: Ikea, borrowed from here).

Creating visualizations such as above, which must be understood pretty much by all (see our post on ‘who’), is a challenge. However, as a rule of thumb, designing visualizations that would be ‘understood by all’ is a good goal, in fact some suggest that a product, such as a visualization, should be designed as if the user is drunk (not an endorsement).

On the other hand, in an exploratory visualization environment, visualization is more of a process, rather than a product. Such an environment ideally gives the users the tools to examine what is in a data set, discover patterns, and gain insights on the studied subject. In an exploratory visualization environment, we don’t necessarily guide the viewer to see certain patterns or emphasize the information we find relevant. Instead, we allow the users as much control as possible, so that they can look at the data from multiple perspectives using various visualization methods, changing the parameters as they see fit (for example, they can reclassify the data with as many different methods and thresholds as they see fit), or bringing in other data sets for comparison. Based on this visual interaction with the underlying data, the user becomes the analyst, and can gain insights and/or come up with some hypotheses. Those hypotheses are typically further tested following the exploratory stage.

Example for ‘explore’: What is going on in the OECD countries?
Discover relationships between social, demographic and economic indicators. Below is an example where we can see some regional statistics by OECD.

ps. Even though documentation can be seen as a form of explanation (“this is what is/was here”), it deserves a footnote. When the goal is documentation, we often try to capture the object/phenomenon of interest as realistically as possible, and ensure that the visualization is reliable to obtain measurements. It is also common to show the object/phenomenon of interest as realistically as possible if the intent is to explore the space or restore a historical artifact. Imagine that the visualization that would be used as a reference in a restoration project, or in some educational contexts where an unknown object/space/time is being studied the first time; recording and viewing the material realistically, for example in stereo, or in virtual reality environments, would allow the best impressions and 3D reconstruction opportunities. In sum, if your goal is to document, consider using photo-realism and capturing of the object of interest with proper/precise geometry.

Last revised: 12th of July 2018, Arzu Çöltekin