Data Visualisation

Data Visualisation is fast and steadily becoming a key strategic approach to data exploration and data analysis within the broader field of BI. As with any other aspect of BI, data visualisation must be clearly defined and understood to be used properly. Considering this, I wanted to provide some thoughts around a few key elements that should always be considered when using data visualisation.

Definition

Data visualisation is the use of visual representations to explore, make sense of and communicate about data. As such, data visualisation is a core and usually essential means to perform data analysis, and then, once the meanings have been discovered and understood, to communicate those meanings to others.

Visualisation provides a very powerful paradigm to make sense of data. By mapping data attributes to visual properties such as position, size, shape and colour, visualisation designers leverage perceptual skills to help users discern and interpret patterns within data. Those patterns are sometimes very hard to detect and analyse when merely looking at the data in a typically tabular textual format.

Purpose

The purpose of data visualisation is to utilise a visual view in order to get to a better understanding of the underlying data and what it represents. Visualisation further allows the information to be put into a business context, thereby allowing for better analyses or better decisions to be made.

More particularly, visual exploratory data analysis is used for:

• Exploration – to find facts of potential interest in a data set.
• Sense-making – to determine what the facts mean (this is also called data analysis or descriptive statistics).
• Narrative – to present specific aspects of the data to others (also called storytelling).
• Monitoring – to detect facts and events of interest and when they occur.
• Prediction – to determine what may happen (also called what-if analysis).

 Graphical representations

It is important to note is that data visualisation tools employ graphical representations of data way beyond the standard graph or chart that you are most likely to picture in your mind based on the description above. Stephen Few, world renowned visualisation expert, uses the term Business Intelligence Visualisation – this testifies that data visualisation encompasses much more than merely displaying raw data graphically. Rather, data visualisation is all about representing and analysing the data in more sophisticated ways. Examples include representing the data in geographical maps, time-series charts, very detailed bar, bullet and fever charts, as well as heat and tree maps. Having a representation of the data in these formats can be considered a more practical way of exploring the data with the objective of finding new insights – all to assist in the representation element of telling the “data story”.

Considerations

Firstly, it must be noted that real data can be very difficult to work with at times and so it must never be mistaken that data visualisation is easy to do purely because it is more graphical. To visualise data it still needs to be understood and analysed effectively. In addition, it takes skill to clearly understand which types of graphs will suit the data under consideration, and which types of graphs will highlight the trends, measures, points or exceptions of interest.

Secondly, data visualisation takes much more than plain coding skills to do well. Creating visuals that offer real meaningful insight to the business requires a broad range of real skills in addition to coding, such as an understanding of graphic design and the factors that affect “good” design when it comes to representing data, together with a deep understanding of the data, what it represents and of course approaches and techniques for doing exploratory data analysis.

Thirdly, to get the most out of data, users must be able to make sense of the visualisations. They must be able to pursue questions, uncover patterns of interest, and identify (and potentially correct) errors. In many cases, visual data analysis requires contextualised human judgments regarding the domain specific significance of the clusters, trends, outliers and other aspects discovered in the data. The business interpretation and contextualisation is crucially important.

There is a lot more to visualisation than drilling through data, drawing graphs and discussing the implications of the findings.  In particular, useful and business-effective visualisations should have the following characteristics:

• They should use a very simple clear representation of the data to give meaningful insight.
• They should represent the data using the correct paradigm (graph type) with the correct dimensions at the correct level of detail.
• They should utilise a very low ink-to-data ratio, to make the data-related message stand out, without being cluttered by meaningless trimmings.

It is therefore very important to know which graphical representations work best with which types of data in order to display the characteristics of importance the best, and in the clearest format.

Concluding remarks

Visualisation can add tremendous business value through exploring, making sense of, and communicating the contents and meaning of the information as presented in the data.

In order to achieve this, you need to utilise a toolset that can effectively access a wide variety of data, and that provides and encourages good visualisations.

In addition, you need the services of a good driver (i.e. a data scientist) that understands the business, that can drive the tool meaningfully and efficiently and that can communicate the findings to the business.

So while data visualisation is highly recommended as a data analysis approach, as well as a paradigm to present findings in the data, BI practitioners should always consider the aspects mentioned above and take the time to understand the concept – as only then will data visualisation become a strategic tool to aid the business in achieving its broader information- and BI-related objectives.