(19-02-2026, 10:37 PM)oshfdk wrote: You are not allowed to view links. Register or Login to view.But why attach a random boolean? As far as I understand, the boolean in your chart is not random, but some metric that depends on the variables.
Exactly! That’s the whole point. The simulation chart I showed you uses a random boolean to demonstrate what happens when the background is independent of the variables being sorted. In the Voynich chart, the boolean (0ed vs ed+) is not random, it is dependent, and that is precisely why the clustering does exist.
(19-02-2026, 10:37 PM)oshfdk wrote: You are not allowed to view links. Register or Login to view.So, it appears this is a simple cutoff filter on one of the variables? What would the random tuples chart show if the background color was selected as A < 5, for example?
That would work, but then you'd be adding a 4th number to the calculation, a constant to act as a filter. My chart only looks at 3 things, if the page has (0 ed or 1 ed in a hapax) or (ed>=2 or in a non hapax) that's the boolean. The normalized count of ho and the count of ed per page are the two variables. There is no numeric cutoff applied to ho or ed density. There is no threshold on the plotted variables. There is no constant added as a filter. Again, I'm no mathematician, but if you can find some functional dependency between a random boolean and the sort key, I'm all ears. That would explain your hypothesis.