4 votesAdminRob Renaud (Programmer/Parsing Monkey/Fake Statistician, Councilroom.com: Dominion Statistics) commented
As far as I can tell, there is no way to actually do this. There is no 'true player identifier' left in the logs. The name is all we've got.
Of course, it's possible to implement merged player pages or allow users to enter in new/old name pairs and use that to do the merge. But that's a lot of effort, compared to you just sticking with your user name ;P.
1 voteAdminRob Renaud (Programmer/Parsing Monkey/Fake Statistician, Councilroom.com: Dominion Statistics) commented
Hi Jonathan. It's currently super simple, as you say. Counts are easy to interpret and easy to update when new data comes in.
I am not actually sure how ANOVA works.
What kind of model would you posit?
Is it possible to online anova, where the cost of updating the results is low for data sets that are small?
10 votesAdminRob Renaud (Programmer/Parsing Monkey/Fake Statistician, Councilroom.com: Dominion Statistics) commented
My problem with the popular buys page is that there is already too much stuff on the page. I think under/overbuying are good first order things to consider if you do poorly with a card. And you can always eye ball the total effect by doing a weighted sum of the with and without. Further you can sort by win rate given avail, which will be super correlated the measure you want already.
17 votesstarted · AdminRob Renaud (Programmer/Parsing Monkey/Fake Statistician, Councilroom.com: Dominion Statistics) responded
I added a deck/supply tracker to the game pages. I think theory’s format makes sense for an editorial, but if you have more control over the display and no human editing, I think the system as implemented now works alright.
The limitation of not knowing what is someone's hand is due to the data just not being present in the isotropic log.
I could certainly organize the information better. As of now, it's still lacking the actual score on a turn, which is the easiest thing to improve.
Yeah, that's mostly what it means.
Also if you are in a 3p game, and neither you nor your opponent win, you'll both get a "tie."
It's there already on the right side, below the graph.
Understanding the graph
Win rate is a generalization of winning probability for multiplayer games.
A win in a 2 player game is worth 2 points, a 3 player game is worth 3, and so on. Someone who plays only 2 player games and wins every time will have a winrate of 2; a 2-player gamer who loses every game has a winrate of 0; a 2-player gamer who wins exactly as often as expected (half wins, half losses) has a winrate of 1.
The win rate of an event is the average number of points for that event. So an event that is uncorrelated with winning will have a win rate of 1.0, and events correlated with winning have a win rate of > 1.0, and losing with < 1.0.
Solo games are filtered out of the database entirely.
Nate/guided/ghorsche did a bit of this already.
Of course, this doesn't preclude me from trying to do it better/always up to take/easier to navigate/etc.
There are a lot of possibilities here, it's just that if I am going to implement something, I want the interface to be nice and usable. I haven't put a lot of thought into the design.
The set of possibly interesting things is very large.
Perhaps encoding the notion of turn intervals into the interface is enough and ranked order (top 5, bottom 5), but it's not clear to me.
If I dedicate some UI to them, then is someone going to want something kind of similar and related, but that doesn't quite fit? Is there some nice unifying concept that I am missing?
On the other hand, it's hard to come up with a simple language to describe what you want
I am not a fan of Microsoft, but I am almost positive that trueskill is going to do a better job than Elo, and Doug already implements it on isotropic.
This is related to two existing ideas:
Even better than just score, if we had a good predictor of actual chance to win, that would be produce even better excitement index.
this graph would be a nice visualization
3 votesstarted · AdminRob Renaud (Programmer/Parsing Monkey/Fake Statistician, Councilroom.com: Dominion Statistics) responded
There overall records are on the player pages, the trueskill is not yet implemented though.
I'll do total record, hopefully tonight. The trueskill stuff will come eventually, but that's a lot more work.
Is this a duplicate of "allow filters on game win pages?"
Do you want to be able to see how well a single card performs when another card (or pair of cards) is in the set?
Some advice on accomplishing this from metaoptimize.