Sydney Data Slam
I took part in Carbon Arts “Sensing Sydney: City Data Slam” – a short but intense experiment in the creative applications of environmental data to better improve public awareness of our environment. The event was part of ISEA 2013.
Here are some images of the event (click on an image for a larger view).
The slam took place at the Object Gallery, Surry Hills, Sydney from 14th – 16th June 2013, 9am – 5pm each day. I worked with Greg Moore working with a dataset of energy consumption for a large collection of Sydney cafes and restaurants. The data was part of an “Energy Efficiency for Small Business Program” provided by the NSW Office of Environment and Heritage.
The data consisted of assessments, down to the individual item level (lights, coffee grinders, refrigerators, etc.), for each cafe participating in the program – the goal being to work out where energy (and hence cost) savings could be made. From the individual item level the data had also been categorised by general technology, such as: Boiler, Compressor, Hotwater, Refrigeration, and so on.
There were 203 individual businesses assessed in the dataset. Here’s a breakdown of total energy costs by postcode:
As can be seen most were in the Sydney city (2000) area. A breakdown by suburb gives similar (but slightly different) information:
Here’s a breakdown of the total energy consumption categorised by technology (lighting, kitchen, cooling, etc.).
A major target of the energy saving campaign was to reduce lighting bills. With lighting the third most expensive energy cost (and relatively easy to reduce), this could mean some savings (the average cost for lighting is around $2.2k per business). However, as the plot below shows there are a range of energy costs for each business.
No information was provided on the size or opening hours of each business in the dataset. So the small number of high-cost businesses on left side might either be very big, open long hours, or very energy inefficient! The pink line shows the mean for all cafes in the dataset.
The main task was to show, for each cafe, where their energy costs were being taken up and to allow for a comparison between other businesses. Here’s a plot with the full total broken down by technology and sorted from the most to the least cost (bottom to top) and most to least total cost (left to right):
Greg developed the colour scheme which allows easy identification of types by association, e.g. refrigeration is blue, heating and hotwater are red. As the earlier plot shows, there is a lot of orange (Kitchen) and blue (Refrigeration) clearly visible for most. There are some anomalies however. Here’s the top 10 most energy-expensive businesses:
For example EESB-00614 seems to be spending a lot on lighting in relation to the others, relative to its use of other technologies. ESSB-00420 has very high HVAC costs.
I developed these visualisations in Mathematica. While each only took a few lines of code (Mathematica uses a very powerful functional programming language), working out the best way to manipulate the data took some time. Our original visualisations that Greg developed on the day used d3.
We also envisaged a number of more advanced methods in the visualisation, such as using geographic information and networking to compare energy consumption between neighbourhood cafes and a campaign to print each cafe’s energy consumption “dan” onto the cups they serve coffee in, so customers could compare easily between cafes and each other on how much energy their cafe uses.
Here’s a sample of our final presentation from Greg:
I also developed some quick 3D form visualisations from the data. The following 3D form shows annual energy use for a single cafe. Each element is a different component of energy use. The colours are muted versions of those used in the 2D visualisations, so the large orange component represents kitchen energy usage.
Notice the very small spike (representing office costs) in the close up below:
Here’s all the cafes in the data set as a forrest of forms:
Here’s a different form that shows all the data as a single form:
Usman Haque (UK)
Jon McCormack (Australia)
Andrea Polli (USA)
Natalie Jeremijenko (USA)
Mitchell Whitelaw (Australia)
Tega Brain (Australia)
Gavin Sade (Australia)
Javier Candeira (Australia)
Pierre Proske (Australia)
Mr Snow (Australia)
Zina Kaye (Australia)
D.V. Rogers (Australia)
Greg Moore (Australia)