Ideas

we can't solve real problem due to only having 15 minute samples, so what can we contribute?

Temporal patterns easily discerned. Rhythmic variance. Patterns in data would emerge. Helpful? We don't know.

Gestalt of a day's data.

1. sample segments, i.e. day by day.

Parameter mapping to pitch accuracy.

Average daily shape.

Average all data from each time segment, sonify variance from average.

Variance from average to pitch?

Zero crossing, pos and neg.

Short auditory marker at crossings?

Listen to different streams, distinguish by pitch area, voice, channel?

Relate power consumption to volume: no consumption, no sound? Combination of gestalt and deviation from average. What sonification can do that graph cannot. Are there patterns of deviation from average?

Could this be predictive of differences between days?

Gestalt of consumption.

Specific temporal events.(review of tools-PD-chop up date first, then load. load and select segments to be played, etc.)

Experimenting

Initial experiment with raw data value to pitch. Add another data channel-data to filter centroid with noise as source.(With given tool, cannot listen to two different techniques, one for each of two data streams.)

(With given tool, cannot use different voice for each data stream without creating, recording, then mixing, i.e. not real time.)

Do the same at different speeds (7, 3, 1 second/s).

Then try different pitch ranges.

Do this for each data channel.

What we found ourselves doing while sub-team worked on average etc.

Do what the tools are good at and see what pops out.

Separate proposal: Calculate averages.

Zero crossings from average is mapped onto 5 different percussion instruments. One for each power consumption consumer data set.

Reflection

• What was the rational behind the approaches? What can we do that shows unique contribution of sonification?

What can we do in short period of time?

Cannot solve real problem, so demonstrate some possibilities?

Data preprocessing may help us find something useful.

• Why did you agree on which ideas to take further?

Can we get this done in short period of time?

Worked with several ideas in parallel. Can they be put together in time?

• What were the key features of the approaches pursued?

1. Simple parameter mapping. Nothing special.
2. Averages, and zero crossings, and deviations.