|
|
|
Thank you very much for the detailed reply. I have tried it several times with different dates and races, but I'm not sure if I'm doing it right.
For example,
http://aligulac.com/api/beta/match/?format=json&date__gte=2013-05-01&date__lt=2013-05-31&rca=Z&rcb=Z&apikey=APIKEY
is supposed to give me the number of ZvZ matches played during May 2013, and it says
meta: {limit: 20, ... total_count: 206} ...
However, list 84 ( http://aligulac.com/periods/84/ ) says that 285 ZvZ games were played between May 2nd and May 15th 2013.
Non-mirror numbers don't seem to match numbers @ http://aligulac.com/reports/balance/ , either.
I'm trying to get these numbers because , frustrated to see nonsense in balance discussion, I am writing an article on current&historical SC2 balance based on Aligulac data among other things. It seems that the number of mirror games is correlated with the balance. (more PvP when P looks OP from other indicators.) So I thought it would be great to include mirror numbers if possible.
|
There's a bug in the inference BBCode output for dual-tournament groups: the entries in the name column are in inverted order, so that the probabilities for the most likely to advance are in the same row as the name of the least likely to advance.
Take e.g. this prediction.
|
On December 21 2013 21:39 Orek wrote:Thank you very much for the detailed reply. I have tried it several times with different dates and races, but I'm not sure if I'm doing it right. For example, http://aligulac.com/api/beta/match/?format=json&date__gte=2013-05-01&date__lt=2013-05-31&rca=Z&rcb=Z&apikey=APIKEYis supposed to give me the number of ZvZ matches played during May 2013, and it says meta: {limit: 20, ... total_count: 206} ... However, list 84 ( http://aligulac.com/periods/84/ ) says that 285 ZvZ games were played between May 2nd and May 15th 2013. Non-mirror numbers don't seem to match numbers @ http://aligulac.com/reports/balance/ , either. I'm trying to get these numbers because , frustrated to see nonsense in balance discussion, I am writing an article on current&historical SC2 balance based on Aligulac data among other things. It seems that the number of mirror games is correlated with the balance. (more PvP when P looks OP from other indicators.) So I thought it would be great to include mirror numbers if possible.
Yes, they won't match. The numbers on /periods/ are the number of games, not matches. Sorry for the confusion!
|
On December 21 2013 22:18 WigglingSquid wrote:There's a bug in the inference BBCode output for dual-tournament groups: the entries in the name column are in inverted order, so that the probabilities for the most likely to advance are in the same row as the name of the least likely to advance. Take e.g. this prediction.
Thank you, fixed!
|
On December 22 2013 02:23 Prillan wrote:Show nested quote +On December 21 2013 21:39 Orek wrote:Thank you very much for the detailed reply. I have tried it several times with different dates and races, but I'm not sure if I'm doing it right. For example, http://aligulac.com/api/beta/match/?format=json&date__gte=2013-05-01&date__lt=2013-05-31&rca=Z&rcb=Z&apikey=APIKEYis supposed to give me the number of ZvZ matches played during May 2013, and it says meta: {limit: 20, ... total_count: 206} ... However, list 84 ( http://aligulac.com/periods/84/ ) says that 285 ZvZ games were played between May 2nd and May 15th 2013. Non-mirror numbers don't seem to match numbers @ http://aligulac.com/reports/balance/ , either. I'm trying to get these numbers because , frustrated to see nonsense in balance discussion, I am writing an article on current&historical SC2 balance based on Aligulac data among other things. It seems that the number of mirror games is correlated with the balance. (more PvP when P looks OP from other indicators.) So I thought it would be great to include mirror numbers if possible. Yes, they won't match. The numbers on /periods/ are the number of games, not matches. Sorry for the confusion!
Does it mean that a single best of 3 series counts as 1 match yet 2 or 3 games?
So, if I understand it correctly, the numbers on /periods/ and balance reports are both games, and then the numbers on that API "total count" are matches?
|
|
|
Denmark3807 Posts
Server down for 20 minutes due to an upgrade (Bitcoin as an earnings type). No worries
|
|
|
Do you count forfeit as a lose ? Didn't know that :o.
|
On December 30 2013 04:36 Faust852 wrote:
Do you count forfeit as a lose ? Didn't know that :o.
A forfeit shouldn't count as a loss, generally it's not included in the database at all. If you see a match in here that wasn't actually played, please point it out so we can correct it!
|
|
|
Interesting. I don't want to delete them too presumptuously since I'm not the one who approved the matches, but if there's no good explanation for including them they'll probably get deleted soon enough.
|
|
|
Some filtering options would be really nice. Specifically, options to filter ratings such that only proleague games are counted, or only foreign vs foreign games, or only Korean vs Korean, but really mostly just interested in ratings in a proleague-only filter 
Also, is it possible to get aligulac to auto-update based on ladder results and provide an aligulac rating for ladder players, say, in gm and masters? It's an interesting idea, but I assume it would be either challenging or impossible depending on what data's available.
|
Denmark3807 Posts
On January 13 2014 03:13 CutTheEnemy wrote:Some filtering options would be really nice. Specifically, options to filter ratings such that only proleague games are counted, or only foreign vs foreign games, or only Korean vs Korean, but really mostly just interested in ratings in a proleague-only filter Also, is it possible to get aligulac to auto-update based on ladder results and provide an aligulac rating for ladder players, say, in gm and masters? It's an interesting idea, but I assume it would be either challenging or impossible depending on what data's available.
You want a rating that ONLY counts proleague games? That would not really be that interesting due to the relative small sample size, anything other than simple win%'s would be more or less useless.
With regards to ladderratings, I'm not sure how interesting this would be:
![[image loading]](http://imgur.com/XL1ppj8.png)
Personally I think the GM ladder shows what you want to see already. Not sure on what types of data can be pulled from battle.net though.
|
On January 13 2014 03:13 CutTheEnemy wrote:Some filtering options would be really nice. Specifically, options to filter ratings such that only proleague games are counted, or only foreign vs foreign games, or only Korean vs Korean, but really mostly just interested in ratings in a proleague-only filter
Recomputing ratings is a nontrivial problem and takes a fair amount of time. If we want to provide such filters we have to account for all the options and have the numbers ready ahead of time. Right now the ratings are refreshed every six hours and it takes about 20-30 minutes each time (maybe about 10-20 of that is in actual rating updates). In addition to that I feel the idea kind of cheapens the idea of the rating system in the first place. The main rating loses some value if people can cherrypick parameters to make it show what they want it to show.
In short, it's doable with some work, but not on a large scale and frankly I'm not so keen on the idea.
|
Liking the rivalries tab! It's really entertaining.and pretty informative on what kind of players certain players play.
|
Noticed a (minor) bug on your reports page. if you clear all 3 plots, the axis is off when you add them again. Screeny: http://i.imgur.com/M48MKny.png
|
Any info on when Aligulac will be back up?
|
|
|
|
|
|
EPT
Maybe they won't be added then.
