You are viewing the MafiaScum.net Wiki. To play the game, visit the forum.
Mafia Statistics Group: Difference between revisions
No edit summary |
No edit summary |
||
| Line 5: | Line 5: | ||
Only completed, Normal games can be included in the data set. Priority for inclusion (as, indeed, inclusion takes time) is given to more recently completed games that are well-characterized in some [http://forum.mafiascum.net/viewtopic.php?f=55&t=37208 archive]. An additional proposed constraint rejects games that feature in bold or voting font words that are close to but not quite "vote" since it will be ambiguous to a naive computer program in these cases whether these phrases implicitly indicative of voting were accepted as votes by a moderator or not. But many complications like that could come up! So instead vote counts might be eyed manually with less of an emphasis on automatizing every little part of the project. | Only completed, Normal games can be included in the data set. Priority for inclusion (as, indeed, inclusion takes time) is given to more recently completed games that are well-characterized in some [http://forum.mafiascum.net/viewtopic.php?f=55&t=37208 archive]. An additional proposed constraint rejects games that feature in bold or voting font words that are close to but not quite "vote" since it will be ambiguous to a naive computer program in these cases whether these phrases implicitly indicative of voting were accepted as votes by a moderator or not. But many complications like that could come up! So instead vote counts might be eyed manually with less of an emphasis on automatizing every little part of the project. | ||
[[Category:Statistics]] | |||
Latest revision as of 05:00, 7 March 2015
The Mafia Statistics Group (MSG) is an informal group devoted to collecting, organizing, and analyzing statistics from Mafia Scum for the sake of understanding the game and the people playing it. It was started by Psyche in 2015, and this page represents the place where the group's work is organized (as opposed to some forum thread). Its current efforts focus on coding game threads to make collecting data concerning them easy in the long run.
Data Set
Statistics is performed on data; the Mafia Statistics Group is thus devoted to amassing a large and robust data set that can be readily treated to statistical analysis. The following protocol dictates its current method for determining which threads to include in its data set:
Only completed, Normal games can be included in the data set. Priority for inclusion (as, indeed, inclusion takes time) is given to more recently completed games that are well-characterized in some archive. An additional proposed constraint rejects games that feature in bold or voting font words that are close to but not quite "vote" since it will be ambiguous to a naive computer program in these cases whether these phrases implicitly indicative of voting were accepted as votes by a moderator or not. But many complications like that could come up! So instead vote counts might be eyed manually with less of an emphasis on automatizing every little part of the project.