http://robowiki.net/w/index.php?action=history&feed=atom&title=Thread%3ATalk%3ADarkcanuck%2FRRServer%2FRatings%2FOutlier_resistant_APS_system%2Freply_%285%29 2026-04-17T23:29:14Z Revision history for this page on the wiki MediaWiki 1.34.1 http://robowiki.net/w/index.php?title=Thread:Talk:Darkcanuck/RRServer/Ratings/Outlier_resistant_APS_system/reply_(5)&diff=23591&oldid=prev 2012-02-16T18:33:47Z

<p>Reply to <a href="/wiki/Thread:Talk:Darkcanuck/RRServer/Ratings/Outlier_resistant_APS_system/reply_(3)" title="Thread:Talk:Darkcanuck/RRServer/Ratings/Outlier resistant APS system/reply (3)">Outlier resistant APS system</a></p> <p><b>New page</b></p><div>About the skewed distributions, fair enough. I still am concerned about the greater noise of medians though.<br /> <br /> The more sophisticated method that was coming to my mind, was calculating the [[wikipedia:Standard score|z-score]] of each sample per pairing, tossing out results that have too extreme of a z-score value, and using the mean of the remaining samples. The reason this appeals to me, is because it changes the existing scoring system as little as possible.<br /> <br /> Most bad results we see are near-zero scores which should be quite distinctly detected by a z-score test, so reliably tossing them out without changing the overall scoring system would be quite doable I think.</div>

Rednaxela