Chronological Path Stability/Uncertainty Analysis

[npcompl33t]

This is a follow-up analysis to my You are not allowed to view links. Register or Login to view. All of the code for this work can be found in the You are not allowed to view links. Register or Login to view..

Thanks for posting these results!

I still disagree that the ordering of sections based on any textual statistics can be interpreted as chronological order.  But let's leave that discussion for other posts.  Here is what those plots tell me about the clustering of pages and bifolios, and your splitting of the usual sections (defined by image topic and A/B language) into sub-sections:

1. I see no evidence in those plots that Herbal 1a and 1b should be separate sub-sections.   
   
2. Ditto for Herbal 3a and 3b. 
   
3. Ditto for Stars 5a and 6.
   
4. Stars 3c and 4c look like they belong together too.
   
5. There are some single pages that look like they don't belong to their assigned sub-sections: maybe two pages each in 1.5a and 2a, one in 3b, two in 5a.  
   

I think you might be misunderstanding the plots. The plots are showing where certain sections may appear across many runs. They are not saying that these sections are shuffled in together in some runs. 

On the contrary, the sections are grouped together and the entire section shifts. In some cases, this effect is so strong that it completely denies content from another section from occupying certain regions, which is what you see with the purple stars section and the bio/yellow stars. Since the above plots combine the positions from all runs, entire sections moving vs individual pages moving look the same. It is not showing what your drawing indicates, and these plots are not even capable of making that distinction, because that was not the goal.

The goal of this post was not to establish these sections move together, I've already done that in the other threads, and i'm happy to continue the debate there. I'm still waiting to see your results running the test you specified with respect to your R^N bet! 

So 3a and 3b may have their orders reversed, but they do not have their pages shuffled together when that happens.  The plot is showing both situations together, which is why they appear overlapping, but it should not be used to estimate group cohesion, because it lacks the relevant information. Again, if you look at the plots in MDS space, it is clear they do form distinct groups.

I did explicitly run a technique designed to identify discontinuous segments of a path. If what you are saying is true, and the sections are not actually connected, or have discontinuous regions between them, then the scEGRET method i used would have detected that. Deleting f58 and some adjacent astro/cosmo pages does cause some of these indicators to activate, but only with certain hyperparameters, so even removing entire segments of the path is not enough to create discontinuous segments. 


Hope that makes sense!