Contact us

Pots (NSC 227 update)

dogmeat
poland

dogmeat

Well-known member
Joined
January 28, 2010
Posts
6,490
Pot 1
Adamsburg
Balearica Island
Denmark of Spears
Fervorosia
Moisantia
Perryfornia
Roseland
Svobodnia
Tamausa & Deltannor
Tanoiro
Xhuxhmaxhuxh
Yeto-Ya-Yeto		Pot 2
Biflovatia
Cherniya
Cydoni-Gibberia
Fierraria
Halito
Illumia
Oosingimaed
Waiting Iist



Pot 3
Elvaci
MatiMati
Reym-L-Dneurb
Serenes
Sodermalm
Tir an Abhainn
Ugaly
Yaponesia



Pot 4
Comino
Dalisska
Effiland
Fervorosia
Kamande
New Acadia
Papendink
Szimbaya Kingdom
Tcher-Racoi
Xochimilia

Pot 5
Begonia
Calypso
Doire
GD Strenci
Insomnea
Konthena
Necluda
New Bander State
Orangualia
Rahasia-Diati

Pot 6
Aimulli
Akatsuki
Belvist
Endore
Kordavian Islands
Marcobia
Redwood Republic
Rumia
Sakuralia
Trollheimr
Vylkuzeme
Zombira

AC2t4oq.png
 
Last edited:
Fearnavigatr

Fearnavigatr

Well-known member
Joined
September 28, 2009
Posts
2,052
Location
Arvidsjaur, Sweden
berlyda said:
I can't tell if this one was newly added or if it's just a coincidence :lol:
Click to expand...
This list is always and undeniably correct for all eternity. Past, present and future. Nothing ever gets added, if you see something that wasn't there before, it's just your mind playing tricks.
 
Last edited:
soundofsilence

soundofsilence

Active member
Joined
April 2, 2023
Posts
718
Pot 3 smokes cannabis
 
soundofsilence

soundofsilence

Active member
Joined
April 2, 2023
Posts
718
How do you make this?
 
soundofsilence

soundofsilence

Active member
Joined
April 2, 2023
Posts
718
How do you make this?
 
dogmeat
poland

dogmeat

Well-known member
Joined
January 28, 2010
Posts
6,490
soundofsilence said:
How do you make this?
Click to expand...
It's based on the voting similarity stats on nscstats.com. The stats are done by a Microsoft Power BI script made by @nofuxCZ (I used to make those manually in Excel before). I copy them into Gephi - free software for creating graphs. The graph is generated by an algorithm that works in a push-pull method. That means, countries repel each other like same poles of two magnets, but at the same time they attract each other with strength equal to their average voting similarity. The circles wobble around for a few seconds, until they reach an equilibrium, where the sum of forces in all directions is zero. Finally, I divide the graph into pots as I see fit.
 
berlyda

berlyda

NSC Mod
Staff member
Joined
September 28, 2009
Posts
4,749
Location
Halito
dogmeat said:
Finally, I divide the graph into pots as I see fit.
Click to expand...
Have you considered a more methodical approach to this step, e.g. k-means clustering?

Each nation's set of voting similarities is essentially a 60-dimensional vector. You could feed these vectors into a k-means clustering algorithm with k=6 and it will group the vectors (i.e. nations) into 6 clusters (i.e. pots) based on minimising distances within each cluster and maximising distances between clusters.

Might be overkill since you're able to do it by eye, but could be a nice way to make it more objective.
 
Paper7
switzerland

Paper7

Well-known member
Joined
May 7, 2011
Posts
4,915
Location
near zurich airport
First i thought: Yeah, Papendink is the new centre of NSC. :D

But then i saw Dogmeats explaination, so it means Papendink is most far away from every other nation in average? ^^
 
dogmeat
poland

dogmeat

Well-known member
Joined
January 28, 2010
Posts
6,490
berlyda said:
Have you considered a more methodical approach to this step, e.g. k-means clustering?

Each nation's set of voting similarities is essentially a 60-dimensional vector. You could feed these vectors into a k-means clustering algorithm with k=6 and it will group the vectors (i.e. nations) into 6 clusters (i.e. pots) based on minimising distances within each cluster and maximising distances between clusters.

Might be overkill since you're able to do it by eye, but could be a nice way to make it more objective.
Click to expand...

I found a Python library with a kmeans function, so it was surprisingly easy, and this is what I got

5 pots:


6 pots:


7 pots:


8 pots:
 
Last edited:
berlyda

berlyda

NSC Mod
Staff member
Joined
September 28, 2009
Posts
4,749
Location
Halito
dogmeat said:
I found a Python library with a kmeans function, so it was surprisingly easy, and this is what I got

5 pots:


6 pots:


7 pots:


8 pots:
Click to expand...
That's awesome :D

Btw I think there are also variations of the algorithm that guarantee the clusters have roughly equal size, but idk if there are nice Python libraries for that.
 
dogmeat
poland

dogmeat

Well-known member
Joined
January 28, 2010
Posts
6,490
Possibly, but I didn't found one. Anyway, I really like this method, and I think I like the 5 pots version the most! Should we use it for NSC?
The biggest con of this method is that we don't get colorful graphs :cry:
 
berlyda

berlyda

NSC Mod
Staff member
Joined
September 28, 2009
Posts
4,749
Location
Halito
dogmeat said:
Possibly, but I didn't found one. Anyway, I really like this method, and I think I like the 5 pots version the most! Should we use it for NSC?
The biggest con of this method is that we don't get colorful graphs :cry:
Click to expand...
I guess you could still generate the graphs using the old method, and then add colours according to the new method?

Or alternatively there are Python libraries for visualising clusters via scatter plots, but you'd need to figure out a good way of representing the 60-dimensional vectors in 2 dimensions.

EDIT: Matplotlib and Seaborn are good visualisation tools if you want to go down that route.
 
Last edited:

Deleted member 5361

Guest
Why don't you make a network graph using networkx or graph-tool? x
 
berlyda

berlyda

NSC Mod
Staff member
Joined
September 28, 2009
Posts
4,749
Location
Halito
berlyda said:
Or alternatively there are Python libraries for visualising clusters via scatter plots, but you'd need to figure out a good way of representing the 60-dimensional vectors in 2 dimensions.
Click to expand...
@dogmeat I just came across a nice way of doing this today, if you're interested:

scikit-learn.org

MDS

Gallery examples: Comparison of Manifold Learning methods Manifold Learning methods on a severed sphere Manifold learning on handwritten digits: Locally Linear Embedding, Isomap… Multi-dimensional ...
scikit-learn.org scikit-learn.org

You can use the MDS class in scikit-learn to generate a 2-dimensional representation of the similarity data in which the distances respect the distances in the original 60-dimensional space. Should have a similar result to your current push-pull method.

Then you can feed the result into a Matplotlib scatter plot:

matplotlib.pyplot.scatter — Matplotlib 3.10.0 documentation

matplotlib.org matplotlib.org

There's a parameter that allows you to colour the points in the scatter plot according to an array of integers; you can use the output of the k-means clustering here and it will give each pot a different colour.

You can also add nation labels to the points in the plot using Matplotlib's annotate() function:

matplotlib.pyplot.annotate — Matplotlib 3.10.0 documentation

matplotlib.org matplotlib.org
 
You must log in or register to reply here.
Top Bottom