Message from @realz
Discord ID: 781067072166952982
Its more edges to do it separately, because I am basically drawing a lot of horizontal lines
(even if the reports are not close together in time)
well now imma try to draw a png via inkscape
this is one of those moments when I wish I had one of those Ryzen processors
anyway, this _better_ result in something that doesn't look like a ball of yarn
I want to see CONSPIRACY
or BUNK
CONPIRACY is if there are clear straight tracks and clear changes to other precincts
at least then we can decide to check this elswhere
and other elections
BUNK is if this happens all over the graph randomly and there is no clear straight lines and switches
oh I just realized something
@DrSammyD I think we should highlight edges where the ratio stays the same, even though there are lots of votes coming in
compared with places where the ratio stays the same but nothing changed in the interim
if I get an actual rendering, I can start doing fancy things like annotating the nodes, and edges in the SVG, and using fancier edge styles to denote kinds of edges
its been generating an svg for 25 mins 😦
@realz you might have said this earlier in the thread above but what are you using for the charting and svg generation. I am a React.JS FE web developer so if you need some JS, XML, HTML, CSS, Json etc looked at for optimizations I'd be happy to give you a hand.
@AntiFish03 I have a git repo with a bunch of other charts if you want to contrib. https://bitcadia.github.io/DownBallot/index.html
I am using python's networkx library to generate dot files, with positions, and then using fdp (graphviz) to generate svg with nodes at those positions
the svgs are themselves ginormous, though chrome seems to handle it a bit
the pngs are ginormous too though
its like handling astronomy data lol
only time I've had such large photos
yea this is the thing, I dunno offhand anything online that can handle graphs this large, except maybe some esoteric webgl graph rendering libs
so I'm not sure if this will be viewable online
I might have to use Gephi (desktop application) to view this
in fact I'll try that now
There are a few especially with an Amazon instance to back it
I don't have access to one at the moment sadly...
first I just need to see if this looks sane
if I get an interesting result (CONSPIRACY or BUNK) then I'll figure a way to show it to everyone
right now I just want to just _see_ it at all lol
LOL
exactly
https://github.com/d3/d3 is a JS library that is designed for heavy duty graphs charting and data visualization
yea I've been using d3
but I don't think its meant to handle 90K-1.5M nodes
and even more edges
You'd ahve to do some massive filtering
yea without knowing what it looks like, I don't know what is safe to filter