Datasets

After much consideration and looking at several different datasets, I have finally decided upon a dataset for the project.

I was thinking about the DJ metaphor quite a little when considering which dataset would be appropriate. DJ'ing involves mixing music, but the underlying factor which must be "correlated" is the rhythm of the music. Rhythm and movement play an important part in the overall DJ setup.

After scouring through the Australian Bureau of Statistics for many hours, I found a particularly interesting dataset that tells the story of the rhythm and movement of people.

The dataset is the "Overseas Arrivals and Departures" within Australia. It is a very complete set of data based on the arrival and departure cards one must fill out when entering or leaving the country.

Some initial thoughts on the use of this dataset:
- considering having the two different "vinyls" being mixed, the arrivals and the departures. (i.e. imagine one vinyl was full of arrival information and one vinyl full of departure information).
- the dataset is very huge - i will have to narrow it down. I was considering using the country specific data - i.e. WHERE people are going to, and WHERE people are coming from.
- Also some thoughts on overlaying this data with data on the general distance from Australia to the countries. This could add more dimension to the data, so you can also see the distances being travelled and by how many people over time.

Visualisation Thoughts
- The user will need some sort of cue as to where within the tracks they are. Normally a DJ will have the visual cue of the record needle, as well as the audio cue of the output music.
- I want the visualisation to be quite minimalistic, perhaps just using lines and the different attributes of a line such as stroke, colour, thickness, direction etc.
- I was thinking what could be interesting would be to map the total amount of people into line thicknesses (such as in Minard's depiction of the fate of Napoleon's army)
- Country locations could also perhaps be used for mapping (though perhaps too literal if it were lines drawn from an imaginary globe...) Maybe if this was abstracted... if australia was the central point of the visualisation, and lines were drawn in the general direction of the country relative to australia?
- Will time be mapped? or will I just let the visualisation change as the user "scratches" through?

Other Thoughts
- How many countries to use? or should I narrow it down to continent totals?
- Should I let the user pick specific countries to display? If so, how would this work?