Day 2 of Dashboard Week saw us provided with the Census of Land Use and Employment (CLUE) provides comprehensive information about land use, employment, and economic activity across the City of Melbourne, making it a valuable research tool.

We were tasked with finding a suitable location for a new cafe and could supplement the data any way we saw fit. I decided to draw on my Economics background and look into proximity to Metro stations and project workforce augmentation in the coming years in Melbourne. I located this data using the Melbourne City data website and was going to approach the task by finding cafes located within a particular radius of a Metro station and in areas of project high economic growth.

The rationale behind this was that (As Tim Hartford state in his excellent book ‘The Undercover Economist) Metro Stations increase foot traffic in areas and are normally people traveling to and from work. This coupled with increased economic activity would provide a solid spot to locate my new cafe. I would have to find. Reading in Yang and Diaz-Rous’s work ‘Walking Distance by Trip Purpose and Population Subgroups’ from the National Library of Medicine in the US, that people found 0.25 miles an acceptable maximum distance to walk for coffee. So that would be my radius around the Metro Stations.

I would therefore want to meet the following criteria.

  • Within radius of 0.4km of a Metro Station
  • Comparatively fewer existing cafes in the area per capita
  • High levels of projected workforce growth

If the location met these three magic requirements, then I could shortlist cafes spots.

Fixin’ with Alteryx’in

I powered up the hyper drive and loaded up the different data sets into Alteryx. Six files in all to combine for the cafe and metro data: the census area and population data and the census areas and business growth data. Alteryx excels at most things, but its spatial tools are always fun to play around with. First up I used the latitude and longitude data to create spatial points for each cafe and metro station. I then created my 0.4km trade area radius around each metro station. A quick select tool to drop unwanted columns and a filter for cafes and then it was spatial match time to find the cafes located withing the 0.4km radius and output to .csv.

Next up I followed the same process for both population and business growth projections. I mapped them to their respective boroughs joined the data to the census area files and outputted as .csv.


That was relatively painless. Now for the next part.

Power BWhy?

I don’t make any secret of my preference for Tableau over PowerBI and when we were told we had to use PowerBI for this dashboard it seemed like a prank at first. PowerBI is fine for click and clack dashboards, but its spatial capacity is woeful. Seen as this was a spatial task it seemed rather odd and just thrown in to make the challenge harder than it needs to be. But as Lord Tennyson once famously stated “Ours is not to wonder why, ours is but to do or die” so, do I did.

I looked into each of the criteria I had set out before and looked at 4 charts.

  1. A Map showing proximity to stations of cafes (bigger bubble = more cafes)
  2. The number of cafes over time in different suburbs
  3. Forecasted population growth over time
  4. Forecasted economic growth over time

Once I had put these together I was able to clearly see that Melbourne CBD is saturated with cafes. Many other areas like South Yarra or West Melbourne did not offer good growth in comparison and therefore I came to my conclusion.

If you are going to open a cafe in Melbourne then it should be in Docklands or Southbank in terms of growth with Southbank edging it in terms of competition being lower. An outsider could be at in North Melbourne, which boasts 3 non-CBD stations and little competition comparatively. Its population forecasts are comparable (indeed better than Docklands) but its growth is more than half that of Southbanks and a third of Docklands by 2040. Parkville offers lower population growth but better economic growth but still lags behind the 2 front runners.

Here is the final dashboard.


The Data School
Author: The Data School