[R] 3D pie

[Peter Ehlers]

This discussion of 3-d pie charts comes at an opportune time. I have
just formulated a new theory of graphical information transfer which
is particularly simple in the case of 3-d pie charts.

Let theta denote the angle between the normal to the pie cylinder and
the pie-eyed line (connecting eye and centre of pie). Then the
information transmitted from pie to viewer is

   K * (pi/2 - theta)^3

for theta in [0, pi/2]. The normalizing constant may be written in
the obvious manner as

   K = 8 * I_0 / pi^3.

I conjecture that I_0 is not large, but I'm still waiting to hear
from Microsoft regarding my application for funding to allow me to
conduct extensive testing.

I'm also working on higher-dimensional generalizations, but even
the 4-d case does not seem to be simple.

Peter Ehlers


Frank E Harrell Jr wrote:

> Rolf Turner wrote:
> 
>>Gabor Grothendieck wrote:
>>
>>
>>
>>>Since everyone else wimped out with a tedious you-do-not-want-to-do-that,
>>>here is a solution that uses R to control Excel and create a 3d chart.
>>
>>    .
>>    .
>>    .
>>
>>People really ***should not*** be encouraged or abetted in
>>wrong-headedness.  Excel is terrible.  Pie charts are terrible.
>>Don't mess with them.  Period.
>>
>>
>>			cheers,
>>
>>				Rolf Turner
>>				rolf at math.unb.ca
> 
> 
> I second that.  Helping people do things known to have major problems 
> with the approaches can actually hurt others in the long run.  2-D pie 
> charts are terrible.  That makes 3-D pie charts terrible to the 3/2 
> power.  Excel has serious errors and is not a good model for 
> reproducible research.