Having announced that I was heading up to Sydney to join Datalicious, I was often asked what I’d miss about Melbourne. Of course, not being able to go to the MCG to watch the footy was certainly high up on the list, as was Saturday mornings spent at the Queen Victoria Market. However, with winter fast approaching, I imagined that Melbourne’s weather was something that I could readily leave behind. Sunny days in Sydney, here we come. But how close was my perception to reality? Was the popularly conceived notion true, that the weather in Sydney is much better than in Melbourne. Time to have a look at the data.

The first step in any decent piece of analysis is to work out what needs to be measured in order to make a valid and reliable judgement of performance, which metrics are important, which metrics are not. As far as the weather goes, the primary and most obvious one is temperature.  Feeling too hot and feeling too cold, seemed to be the most common complaint from people. It's also important to consider that according to the Bureau of Meteorology, how people ‘feel’ the temperature is also dependent on how windy it is, the humidity and if the sun is out. Finally, whether it is raining, how much, and for how long, also seemed to me to be a major dimension of people’s judgement of the weather.

OK, now that we’ve established the important measures, we next have to see what data is available and how the metrics might be defined. Luckily, the Bureau of Meteorology are equally as nerdy as myself and have measured all these things over the past 150 years and they have made the past 12 months of daily data available to download from their website. This is how they are defined:

+ Min Temperature - minimum temperature in the 24 hours to 9am

+ Max Temperature - maximum temperature in the 24 hours from 9am

+ Rain - precipitation in the 24 hours to 9am

+ Sunshine - bright sunshine in the 24 hours to midnight

+ Humidity - relative humidity at 9 am/3 pm

+ Wind speed averaged over 10 minutes prior to 9am/3pm

I’ve then put this data into the Tableau visualisation engine - below are two screen shots of the reports I created but please also check out the interactive Sydney vs. Melbourne Weather dashboard on Tableau Public and change the definition of good weather!

While I can now easily describe and ‘see’ the weather for Sydney and Melbourne, I can’t yet test my hypothesis because it includes a normative term: ‘better’. This is the difficult part. What is the benchmark for success? Is there a universally acceptable definition of good or bad weather? I doubt it. In the absence of such a universal definition, I’m just going to go with my own:

+ Few days with temperatures below 20 and few with temperatures above 30

+ Few days with o’night temperatures below 10

+ Few instances of consecutive days of rain (a rainy day is 5mm more of rain)

+ At least 1 in 2 days of sunshine (a sunny day is 8+ hours of bright sunshine)

+ Low wind speeds regardless of temperature

+ Lower than 80% humidity (when it’s not already raining)

If you look at the charts, Sydney wins on most of these measures with daytime temperatures in a better range, no cold nights, lower wind speeds (especially at temperature extremes), more sunny days, and more consecutive sunny days. It does rain more frequently in Sydney and the humidity is higher, but it’s tolerable. Sydney does have better weather then Melbourne. Case closed.

Now, I wonder which city has worse traffic congestion … let me know if you’ve got any data!