3D Human Services Data Hub

Introduction

This visualisation shows information about similar LGAs (Local Government Areas), in terms of:

• Population
• Socio-economic wellness (SEIFA)
• Number of Govt Service Contracts

Geographic information is traditionally displayed on a map. That choice of spatial representation limits an ability to notice similarities. This example shows another way to think about the data. It shows the LGAs as "dots in a box". The position of each dot (LGA) is based on its Population, SEIFA, and contracts. If the LGAs are similar enough, they will form a cluster (shown by the transparent grey isosurface). The colours of the dots are actually also representing SEIFA (Blue to Red). It's reduncant, since SEIFA is already an axis. Colour could have been used to show a 4th variable, like the type of service, or the Govt Agency (there's four).

Other users have projected the colour of the dots out to the isosurface, so instead of grey, the clusters take on the dot's colours.

Background

This visualisation builds on the work done in the 2D Human Services Data Hub project. The motivation for this alternate representation arose from the work that was done using Google Earth. The problem with the Google Earth visualisation was the population data was displayed by local government area. Rather than colouring the local government areas by population, we displayed this information using extruded 3D towers.

These towers took the shape of the local government area and the height was used to represent the size of the population in that region. The differing sizes of the local government areas however, meant that some of the 3D towers appeared greater in volume than the others. Unless inspected carefully, these differences in volume could mislead a viewer by suggested a greater population than that which was actually present.

This type of visual exaggeration can easily result in a misrepresentation of the underlying data – in terms of accurate data science, this is an outcome to be avoided. In order to resolve this problem and provide a more accurate representation of the information, we developed this alternative 3D visualisation of the same data set.

Cluster Bubbles

This visualisation contains a 3D graph, where each axis of the graph (or box) represents one of the variables being plotted. These variables are number of social housing contracts, SEIFA (Socio-Economic Index For Areas) and population. Each point in the graph corresponds to a particular local government area.

SEIFA is not an indicator of a rich or poor area, it’s more a measurement of the conditions of the area e.g. access to schools, public transport etc. The number displayed in brackets beside each LGA represents the population of the area. The other value (yellow) is the total number of Govt Contracts for the LGA.

Clear grey metaballs are placed at each LGA. If close enough to overlap, they join like water drops to make a surface. Sometimes called an "iso-surface" these blobs show similar places - LGAs with similar population, socio-economics, and Government Service.

An interesting example of this behaviour is visible in the screen shot above – the characteristics of Ryde mean that it is almost close enough to join Randwick and Hornsby to form a single large cluster. However, a small change in the data for Ryde would produce a large difference in the iso-surface shape. It could fully join the blob, or might separate.

Typically geographic data is shown on a map. Data for Tamworth is plotted where Tamworth appears in NSW. If you already know where Tamworth is, then perhaps there's a better use for the spatial representation? These clusters indicate similarities. Closer investigation shows some unexpected group members worth further investigation, outliers, similar outliers, a general trend for pairs of axes, gaps in service provision, and a few special cases.

Why is this important?

Normally, visualisations are used to find answers but in this situation it actually raises questions.

In the 2D visualisation of this information, we were looking at a map: the map is really just telling you where Ryde is located. If you already know where Ryde is located, in some ways you don’t really need the map -it’s just telling you where it is and what the numbers are. In the 3D visualisation however, the local government areas are positioned spatially based on the parameter space rather than geography.

In the visualisation, Parramatta and Newcastle are considered similar. This is interesting because they are not near each other geographically but they are considered similar in terms of contracts, population and SEIFA index.