Initial Ideas

• Calling for Beacons, Markers
• Seems very periodic
• maybe difference and ratio are more important than the curves themselves
• Make streams for the 5 components - blending for sum.

Interactivity for emphasizing single components

• Co-inciding peaks are crucial information
• Continous cues for the time instead of clicks (e.g. Noise for daylight)
• Using localisation (5 levels already)

Experimenting

• First try with the Sketch - sine waves -
• channels map onto amplitude, losing a lot... (compensating for frequencies, exponential)
• splitting up channels ? - FM synthesis
• trying differences of 2 channels (diff and ratio)
• merging pitch mapping with amplitudes

Evaluation

Not significant enough for decisions - not very convincing

Better Aesthetics would help

Predictions vs. Analysis

What next?

Compare to existing predictions in the domain

Work with cost values - patterns should be reflected

Real data - at higher sampling rates

Compare to previous day

Interactive Sonification

Discrete notes - play as an instrument