Author: Jon Dart
Date: 07:11:02 04/05/04
Go up one level in this thread
> So, what do you think? Anybody heard of such a project before? Do you all >think it would work? I'd be interested in any thoughts you have. Andrew I haven't heard of this being applied to chess, but I taught university-level statistics for three years, so I know a bit about it ;-). Regression assumes continuously variable quantities (real numbers) so it probably isn't a good choice for finding the strength of relationship between chess position features and game outcome (the features you mention have integer values). If the features you are comparing have a few mutually exclusive values, you could use the chi-square test to see if they are significant. Otherwise, you'd probably need a non-parameteric test. These tests could at least tell you if the relationship between feature and outcome is significant. They won't really help you analyze complex relationships among features that contribute to the outcome. You'd need somthing like cluster analysis for that. --Jon
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.