You've probably solved this and moved on
Actually no. This is a backburner project I've been revisiting in quiet moments on and for several years. I've arrived at several partial solutions (to the underlying problem rather than the graphing), but I'm generally looking for faster, better, more complete solutions and the graphing is means to one possibility.
I increasingly feel a good candidate to represent a large volume of data like this would be a heatmap.
Heatmaps work well for 3D data on a 2D graph per the examples you linked. They can also work pretty well for 4D data on a 3D graph where the 4th dimension is a continuous function  like this for example.
But for my datasets, the 4th dimension is discrete and can cover a huge range  potentially more than there are colors in 24bits. But even when the range is more reasonable, picking out small variations in the colors of individual pixels is beyond the scope of human eyes. Mine at least.
The best solution I've thought of to date, would be to plot the 4th dimension as a vertical line on top of the 3rd dimension point. Piss poor description :(
If you imagine the following to be a 2D slice through a 3D plot which is oriented squarely with X running horizontally, Y running 'into' the screen, and Z vertically. Then the vertical lines represent the (log10 of the) 4th dimension with the base of the line being the Z point:
Zmax 
 
   
   
   
    
      
       
       
      
    
    
    
  

Zmin ++
Xmin Xmax
The vertical lines would be continuous. Their low extremity would be the Zvalue. Their length the log10 of the frequency.
Now put many of those slices together in the Y dimension with perspective etc. in a 3D plot and you get a visual representation that shows how the data clusters, but also the peaks and outliers within each cluster.
My only problem now is that I cannot work out how to persuade gnuplot to draw it for me :(
With the rise and rise of 'Social' network sites: 'Computers are making people easier to use everyday'
Examine what is said, not who speaks  Silence betokens consent  Love the truth but pardon error.
"Science is about questioning the status quo. Questioning authority".
