Life In 19x19 http://prod.lifein19x19.com/ |
|
A histogram of game lengths and other play number based stat http://prod.lifein19x19.com/viewtopic.php?f=10&t=8720 |
Page 1 of 1 |
Author: | SmoothOper [ Mon Jul 15, 2013 8:25 am ] |
Post subject: | A histogram of game lengths and other play number based stat |
Is there anyway to get a histogram of the number of plays per-game maybe in Kombilo, maybe also broken down by opening? Along these lines, is there a way to get histograms on the play number when certain events, such as corner invasions, occur. And for end game tesuji, how many moves from the end do tesuji, such as monkey jumps, occur. I am looking for not only the pattern but when it happens, relative to either the start or the end of the game. I don't think resigning should matter so much in this type of analysis. |
Author: | snorri [ Tue Jul 16, 2013 8:07 pm ] |
Post subject: | Re: A histogram of game lengths and other play number based |
SmoothOper wrote: Is there anyway to get a histogram of the number of plays per-game maybe in Kombilo, maybe also broken down by opening? Along these lines, is there a way to get histograms on the play number when certain events, such as corner invasions, occur. And for end game tesuji, how many moves from the end do tesuji, such as monkey jumps, occur. I am looking for not only the pattern but when it happens, relative to either the start or the end of the game. I don't think resigning should matter so much in this type of analysis. The answer to some of these questions is yes if you are willing to write some code. See the scripting documentation. For example, for your 1st question, I am able to get this kind of data with a simple (though inefficient) script: Attachment: gamelengthhistogram.png [ 20.41 KiB | Viewed 3148 times ] I could easily break that down by opening, but it would require some kind of definition of what you mean by opening. If there is a specific opening you care about I can regenerate. For example, here is the data restricted to games with black sanrensei vs. white nirensei and the classic Shasaku 1-3-5: Attachment: movebyopenings.png [ 24.49 KiB | Viewed 3135 times ] So those two openings are about as different as can be, but it doesn't really affect game length much. However, for some of your questions, I don't know if it's very easy or might require you digging into the C++ code of libkombilo. However, the Kombilo author is on L19. You could PM ugoetz and see if he can assist. |
Author: | SmoothOper [ Wed Jul 17, 2013 11:02 am ] |
Post subject: | Re: A histogram of game lengths and other play number based |
snorri wrote: SmoothOper wrote: Is there anyway to get a histogram of the number of plays per-game maybe in Kombilo, maybe also broken down by opening? Along these lines, is there a way to get histograms on the play number when certain events, such as corner invasions, occur. And for end game tesuji, how many moves from the end do tesuji, such as monkey jumps, occur. I am looking for not only the pattern but when it happens, relative to either the start or the end of the game. I don't think resigning should matter so much in this type of analysis. The answer to some of these questions is yes if you are willing to write some code. See the scripting documentation. For example, for your 1st question, I am able to get this kind of data with a simple (though inefficient) script: Attachment: gamelengthhistogram.png I could easily break that down by opening, but it would require some kind of definition of what you mean by opening. If there is a specific opening you care about I can regenerate. For example, here is the data restricted to games with black sanrensei vs. white nirensei and the classic Shasaku 1-3-5: Attachment: movebyopenings.png So those two openings are about as different as can be, but it doesn't really affect game length much. However, for some of your questions, I don't know if it's very easy or might require you digging into the C++ code of libkombilo. However, the Kombilo author is on L19. You could PM ugoetz and see if he can assist. Thanks snorri, that is really cool! Do you mind if I ask about your script? I am good at C++, but my python is rudimentary. So there is sort of a two peak distribution in game length. I wonder what accounts for that, early resigns? It seems Shusaku has somewhat shorter games overall, especially if you consider the number of games in the 140 move range and in the 280 move range, it doesn't look like the histograms are standardized though. |
Author: | walleye [ Sun Jul 21, 2013 4:46 pm ] |
Post subject: | Re: A histogram of game lengths and other play number based |
Could you make a plot using only the games that reached the scoring phase? Excluding the game that ended by resign or were left unfinished. Thanks. |
Page 1 of 1 | All times are UTC - 8 hours [ DST ] |
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group http://www.phpbb.com/ |