A good visualization vehicle is a power distribution graphic for an electronics system.

Input Power is a process whose “weight” is proportional to the power draw at that stage. (Power Transmitted + Power Consumed). Subsequent stages draw from the input stage with the same transmission/consumption rules.

An algorithm that would drop process shapes as nodes in the visio canvas would be required. A spreadsheet would drive the power parameters that determine the relative “weight” of the sankey edge connections.

The drawing would benefit from bending of the sankey connectors out to the star nodes. If the connectors new the power consumer connection points, then the algorithm would need to connect them nicely to the source node so that they don’t overlap.

Since the power is finite and well determined, the arclength of a source node should nicely match the collective widths of the consumer nodes normalized to avoide the full 360 degree outburst.

It seems that with a smart arrow and some help from the elegent circular pattern smart shapes that all that is lacking is a “NodeXL” add-in for visio, “NodeVz.”

Thanks,

Nick

While I may have built this manually in a past life, now that I’ve discovered your cool pre-builts I now expect my every whim to have been anticipated and and incorporated into a nice little package.

I agree, this isn’t a solution looking for a problem, I have that exact problem and I would LOVE a solution like this.

Dahn

For instance, if we have two arrows or paths of the same thickness, the outer one might look bigger because it has a longer path and therefore more area.”
——————————————

In a recursive process like, say, stellar nuclear reactions or the Krebs cycle, you have to think carefully about how you use area, and you might add some novelty and cleverness to the “state of the art” here.

Sankey diagrams use line thickness to represent quantities– in recursive processes, the process quantities are going to be tiny compared to the available feedstock (eg, nuclear reactions in the sun will take 6 billion years to convert the sun’s hydrogen supply to helium). So if you did everything to scale, you’d either get a vanishingly thin line (for helium output relative to hydrogen stock)– or you’d create something misleading.

I think it would be interesting to explore the the idea of representing the central cycle without scale, but use the line thickness for the scale of the inputs and the outputs.

It bears some thinking about, and I think the best answer will depend on the specific process one is trying to illustrate, and their scale.

There is no obvious “right answer” to your question, not to me, anyway.

Do you think the circular outlay would cause any misrepresentation in the graphic?

For instance, if we have two arrows or paths of the same thickness, the outer one might look bigger because it has a longer path and therefore more area.

I suppose outer paths could be made shorter using automated calculations.

There are so many bad graphics in the world that make misleading comparisons (like elliptical pie charts!) We don’t want the guys at Junk Charts or FlowingData to jump down our throats!

The Bethe-Weizsacker cycle (or “C-N-O” Cycle) for example
http://en.wikipedia.org/wiki/CNO_cycle

This and the previous three that I mentioned above more or less demand a circular approach. Unlike Napoleon’s Army in Russia, there’s no “beginning” and “end” — these cycles are essentially continuous (until they exhaust fuel or whatever), so the the Minard-style Sankey diagram is less appropriate than a circular one.

A few potential applications come to mind.

The most obvious are recursive industrial processes, where outputs are fed back into the system. Engines (turbocharged, turbofan) come to mind– what one is trying to understand in analyzing a system like this is how performance varies with the degree of feedback. Cogeneration also roughly fits this paradigm.

In financial models, the radial paradigm would be a neat way to represent reinvestment

There are a lot of biological processes which are described with radial paradigms– the most celebrated is probably the Krebs’ Cycle, but there are lots more.

One could also depict information based on the arc-length of an arrow (like pie charts do with sector-size). Distance from center could also have meaning.

One potential problem is that as you move out from the center, arrows of the same thickness and arc-length will have a lot more area further from the center than nearer. Such perceived “bigness” can be a problem when trying to accurately depict information.

- Chris

They could start with plan card stock and over 10 steps like printing, trimming, etc arrive at the final pack. At each stage of the process they had some wastage and they would track this wastage regularly. Similar situations exist in many industries and it is good to know where everything is going.

Also from a financial analysis perspective financial analysts want to know what is being spent on various aspects of a business. We would do this using a bar graph or a pie chart. It would be a great idea to see using circular sankeys. I think the width of the arrow is a better visualization of the amount being spent rather than a slice of a pie.

Regards

Tangents would be good for a process that is really rotating.

g