This project explores the correlation between poker machines and socio-economic status in Sydney.

Responsive image

I worked on this project while

interning at Small Multiples.



Data Processing

Data Visualisation


June 2019


Resonsive image QGIS

Resonsive image Carto

Resonsive image Kepler


I began by collecting data for every liquor and gambling club in Sydney that had poker machines. This included information about how many machines they had and in what locations. I then geocoded each club so I could plot them on a map.

To find out whether the number of poker machines per club had a correlation with a suburbs socio-economic status, I needed overlay this information on top of Socio-Economic Indexes for Areas (SEIFA) data. In order to do this I obtained the shapefile of suburbs in NSW and merged them with this data.

One great application to help me do this was QGIS, a tool to create, visualise and analyse geospatial data. I used this to visualise the range of advantaged and disadvantaged suburbs in NSW. I imported the shapefile and combined it with the SEIFA data to make a visible choropleth that I could lay it against my other data.

Responsive image
Responsive image

I used an online map visualising tool called Carto to quickly plot the number of poker machines that were at each club. The larger and darker the circle, the more machines. Initially it was not clear whether there was something significant being shown here. There were so many points on a map, that were scattered all over Sydney.

It wasn’t until I added the choropleth layer that the results became interesting. The darker red, or more disadvantaged suburbs, had a larger cluster of points. This showed that the more socio-economically disadvantaged suburbs do actually have a larger number of poker machines in the area.

Responsive image
Responsive image
Plot of number of poker machines per club

This data was interesting, but how could it be made more meaningful? To get a more representative number I had to look at the population of each area. I collected data of the net profit of each club in their Local Governement Area (LGA), as there was not this data avaliable for each suburb. This meant that it would not work with the shapefile I already processed and I would need to make another one for LGA’s. I went back into QGIS, gathered the new data and combined each LGA with their respective SEIFA score.


I had the datasets, and now I wanted to see what existing map visualisations were out there. I wanted to discover what design choices people made and how the interaction could be enhanced. Color Brewer was a great resource for what color schemes to follow, and how they could be used to help distinguish different regions from each other. It was important to be able to view all like colours on a screen without confusing them with each other. Most maps I looked at had a choropleth visualisation, others had only points on a map, and I needed to combine both.

After researching several interactive visualisations, I decided upon the best way I could visualize my data on a map and the colour schemes I could use. Reds and oranges were used when representing negative or dangerous situations. Pink was used in the representation of lighter-hearted topics. Blue and green were commonly used in social or economic related maps. They all communicate their messages effectively.


Once the data had been finalised and I had thought about how I was going to visualise the maps, I got into Carto again and started creating more.

Responsive image
Responsive image
Responsive image
Base Map
Data Points

The first step was finding the best base map. The base map needed to be a neutral colour to lay the foundation of the choropleth that would be on top. Carto provided a few default colours of which I chose the grey one as no other colours were really necessary.

The next step was adding the shapefile data on top. The blue colours represent the socio-economic status - the darker the area, the more disadvantaged.

Finally, I added the poker machine data as points. One problem I came across was the accuracy of the location of the dots, some of them had to be moved manually because the API I used had found slightly incorrect coordinates for some areas. I looked online for the coordinates of the inaccurate areas and edited the excel file accordingly.

After making the maps in Carto, I explored other tools online to achieve a similar goal. A great one I found for easily making and sharing maps is Kepler, a program created by Uber to help people create interactive map visualisations.

Machines Per Capita
Net Profit Per Capita
Net Profit Per Poker Machine