K-Means Clustering: Unsupervised Learning Applied on Magic:The Gathering | Hacker News Reader

K-Means Clustering: Unsupervised Learning Applied on Magic:The Gathering

datastuff.tech

57 pointsstrikingloo18 comments

> I just didn’t want to mix my M:tG findings with this tutorial so that readers who are into Data Science but not into the game won’t be bored.

I'd encourage you to add some examples here, even if they're dumbed down. Without that, the article is not telling me what's been achieved through the process.

It's visible on the notebook, but I will add them to the article if more people insist. I'm still kinda new to writing, and wasn't sure if it would make the article better or worse. As a summary, I can tel you one of the clusters learned all about elves and green cards, there was a cluster that played Tron, and so on.

I think you should add them, personally.

a picture says a thousand words, I was hoping to see the clusters with discussion of your findings.

Awesome! I'll work on that side of the article then, I'll probably have it ready by tomorrow!

Right, this is just a recommendation to help you improve the article format. As of now, I see “I applied k-means clustering, you won’t believe what happened next”.

I want to see what happened next, but working through an ipython notebook is a big ask, heh

Unless I'm blind, I don't actually see the notebook checked into the GitHub repo.

https://github.com/StrikingLoo/MtGRecommender/blob/master/mt... Here you go!

Here you go! http://www.datastuff.tech/data-analysis/magic-the-gathering-...

Neat idea, but I'm not sure the approach of using euclidian distance on what's essentially a categorical variable is valid. Instead try a different clustering algorithm like K-prototypes [1], or Gower distance instead of euclidian.

[1] https://pdfs.semanticscholar.org/d42b/b5ad2d03be6d8fefa63d25...

Edit: Thinking about it more, you could treat the cards in each deck as a bag of words and run LDA on it. Alternatively create an embedding (just keep in mind skip-grams aren't meaningful for decks of cards) and cluster those.

I like the LDA one (https://towardsdatascience.com/finding-magic-the-gathering-a...) using non parametric bayesian.

But seeing different cluster algorithms in action is very enlightening.

How many people would like to play MtG vs a good AI?

How dominant is the social aspect when playing online?

(Lets ignore copy right issues for the moment)

I would love to play MtG against a decent—and particularly, scalable—AI.

Not only am I not all that good at the game, I have been burned too many times by bad online play experiences—between people trolling, people at vastly higher levels of play than me, and people who are just rude and mean—to really want to play against real people I don't know most of the time.

But I do like playing. So an AI that could provide a reasonable challenge would be ideal.

don’t be discouraged by the trolls. Do go to local, in person, events (Friday Night Magic) as real life players are among the nicest and most supportive people I’ve seen.

Unfortunately, the closest Friendly Local Game Store is 40 minutes' drive away. I've been to a few events there (prereleases, mostly), and they're OK, but not worth the trip for me.

Have you played on Arena? I'm struggling to imagine how anyone could meaningfully troll you on that apart from by spamming the 6 pre-canned communications (which you can mute).

Sadly, no. It doesn't run on Macs, and I don't currently have the funds to justify a second computer just for Windows-only games.

a lot of people actually. MtG AI is notoriously hard to implement because of the huge number of abilities and the evolving nature of the game.

Reply on news.ycombinator.com