Density is a factor for Covid-19 growth in NSW but Age group analysis shows some signs of hope
With a quieter than normal long weekend we’re experiencing, I thought I would take a look at adding some further nuances to the NSW Covid-19 tracker (available here: LINK)
This includes demographic data for LGA’s (Local Government Areas) which means I can look at population size, density and age cohorts as a factor for case growth.
As well as an interactive map showcasing the spread.
Population, Density and Age Brackets
With all the data that’s out there across multiple news and media websites, none showcase how population factors might have an effect on this virus. Whilst the data I have gathered does not pertain to population density, larger populations can give some indication of that.
To get population data we go to the Australian Bureau of Statistics (ABS) and it’s Historical Population series for 2019 (https://www.abs.gov.au/AUSSTATS/abs@.nsf/DetailsPage/3105.0.65.0012016?OpenDocument). I have pulled in this information at an LGA level as the suburb data as I’ve not yet found a source for individual suburb population (though could add this later on).
I’ve also used data from this ABS table (https://www.abs.gov.au/AUSSTATS/abs@.nsf/DetailsPage/3235.02018?OpenDocument) to get details on age cohorts, specifically percentages of 15–64 year old persons in a suburb as well as percentages of those over 65 years old.
Population size
Using this information I created a scatter chart which has population on the Y-axis and cumulative cases (up to 11th April) on the X-Axis.
From the data we can see that LGA’s like Sutherland, Northern Beaches and Sydney fall in line with their expectation of being larger populations and also showcasing higher numbers of infections. However, Waverley, the known hotspot across NSW is not as highly populated yet outranks all other suburbs in terms of confirmed cases. Put this down to poor social distancing on the part of its locals. On the other hand, we see Canterbury-Bankstown and Blacktown, both LGA’s with higher population density yet middle of the pack in terms of confirmed cases.
Additionally, I have added a slider filter to this visualisation that means the users can filter out suburbs by size.
So, for example, we can look at LGA’s with populations over 100,000
Or, we can look at smaller LGA’s with 50,000 or less persons (this one should worry me as I live near one of the outliers, Mosman).
Population Density
Density = Mass / Volume
The above formula is how density is measured and one of the topics spoken about with regards to Covid-19 is whether more densely populated areas are prone to higher numbers of confirmed cases. It’s fun to speculate but I’d rather look at the numbers.
Since we have population numbers for the various LGA’s we can use that as the ‘mass’ part of the equation. For ‘volume’ I found an ABS dataset that shows the area size in square kilometers for each LGA (https://www.abs.gov.au/AUSSTATS/abs@.nsf/DetailsPage/1270.0.55.003July%202018?OpenDocument). It’s based on the ‘Albers Conic Equal Area projection’ which you can read about here if you’re interested (LINK). Bringing these 2 together I created a density calculation of Persons per Square Kilometer.
Plotting this on a map against cumulative cases shows a positive slope. This means that more densely populated areas are the ones seeing higher case numbers. As expected, and confirmed by the numbers.
Age cohorts
The LGA data I collected also contained age cohort details (https://www.abs.gov.au/AUSSTATS/abs@.nsf/DetailsPage/3235.02018?OpenDocument), specifically in the categories of percentage of persons in the 15–64 year age bracket and those over 65 years old. We can also plot these 2 factors on our graph to gleam further insights.
For the 15–64 year old age group we see a positive sloping trend. This makes sense as the population and density graphs align with people in this group more likely being those who live with many others and are in suburbs where housing is more closely packed together.
For those over 65 the trend is negative. This is a good thing as it means there is a much weaker correlation between LGA’s where older residents are more likely to be and the spread of this virus.
As we’ve seen in the global data on Covid-19, the virus has higher fatality rates amongst older persons, the fact that the main areas in NSW being affected are not the ones where most of those older residents live and this is a positive sign.
Map
As a reminder, the NSW data comes from NSW Open Data (http://data.nsw.gov.au/nsw-covid-19-data). It has some location data but only lists suburbs by postcodes and has no latitude and longitude data that’s necessary to do mapping.
Scouring the web for this data (at least for NSW) I stumbled across the website of Matthew Proctor who provides a service to showcase this info: https://www.matthewproctor.com/full_australian_postcodes_nsw. He collects and displays this because Australia Post used to provide this information for free but has since removed it.
Thankfully Matt’s site has this information and it means we can see the spread over time like this
5th March, 2020
19th March, 2020
2nd April, 2020
9th April, 2020
Contact details
Stay tuned for more updates like this in future and if you would like to get in touch with me about this or other posts on this blog then feel free to reach out below:
Mark Monfort (Head of Data Analytics and Technology)
- Phone: 02 8262 8700
- Email: mmonfort@prosperity.com.au
