The four levels of data visualisation
Professor Chen and Amos Golan, a Senior Associate, at Pembroke College, Oxford, identified four levels of visualisation in their 2015 paper ‘What may Visualisation Processes Optimize?’. This extract summarises the four levels of visualisation.
Level 1: Disseminative Visualisation
Visualisation is a presentational aid for disseminating information or insight to others. The analyst who created the visualisation does not have a question about the data, focusing primarily on informing others: “This is A!” where A may be a fact, a piece of information, an understanding, etc.
For example, Ed’s spiral, which you saw in Step 1.7 is an engaging way to see how global average temperature is changing.
Level 2: Observational Visualisation
Visualisation is an operational aid that enables intuitive and/or speedy observation of captured data. It is often a part of the routine operations of an analyst, and the questions to be answered may typically be in the forms of “What has happened?” “When and where A, B, C, etc., happened?”
For example, monitoring a stream of data from a live sensor measuring air temperature or a ‘traffic flow’ which is a dataset used as part of a project with Highways England. You’ll hear more about this project in 3.12.
Level 3: Analytical Visualisation
Visualisation is an investigative aid for examining and understanding complex relationships (e.g., correlation, association, causality, contradiction, and so on). The questions to be answered are typically in the forms of “What does A relate to?” and “Why?”
For example, Benefits of Urban Nature to You (BOUNTY), a case study you will explore in the next Step, shows how changing one thing in a neighbourhood, such as the green space, can affect the flood risk, air quality and house prices simultaneously.
Level 4: Model-developmental Visualisation
Visualisation is a developmental aid for improving existing models, methods, algorithms and systems, as well as for creating new ones. The questions to be answered are typically in the forms of “How does A lead to B?” and “What are the exact steps from A to B?”
Visualisation is used to analyse and diagnose complex simulations of the Earth system (eg atmosphere, oceans), to see if the simulation is producing reasonable results. Based on this, the simulation code may be modified and improved, perhaps to help it simulate more physical processes to increase the realism of the model.
If you’re keen to find out more about the levels of visualisation and how to identify the level then you may like to read ‘What may Visualisation Processes Optimize?’. Please note: this paper is behind a pay-wall.
In the next few Steps, you’ll explore several data science projects where visualisations have been integral to their success. Don’t forget to mark this Step as complete before you move on.