Skip to main content
3 events
when toggle format what by license comment
Jun 12, 2014 at 7:18 comment added andy256 Yes; my statistics is not strong enough to take the argument about expected distributions further. As you say, a visual representation (or several) of the data would help with understanding it, and hence with producing a viable algorithm.
Jun 12, 2014 at 6:53 comment added Christian Sauer I like this. Also, I would rather ask this on CrossValidated. I would add standard deviation and eventually a significance test to determine which cases are unusual (under- or overperforming). A boxplot might be a great way to visualize the distribution
Jun 12, 2014 at 2:55 history answered andy256 CC BY-SA 3.0