Quote:
Originally Posted by BMSOTM
How come 1257 is #2 on the low goal list, ahead of 708, but not even top 5 for boulder volume when 708 didn't score high? (or at least, not that I saw) Not that I take issue with it, data is data, but I'm genuinely curious how this result is possible.
|
Thanks for asking! Yeah, you can also see a similar effect with 5895 coming in ahead of 3314 in overall boulder volume but behind in teleop high goals.
When I was working on it initially, I realized that consistency was overvalued in the standard error statistic*, and a team that was consistently slightly better than the average team came out far ahead of everyone else, even those that had small inconsistencies but were generally better (significantly higher average.) This is because using the t-distribution isn't precisely telling us how good a team is, but rather how unlikely their performance is given that we assume that they are the average team.
The way I solved this is by restricting analyses of individual fields to a select set of teams rather than all teams at the competition, which raised the average comparatively and reduced the overvaluing of absurdly consistent teams. I kept the ones with all teams analyzed, but honestly reality-checking the latter made me realize that restricting the number of teams for more specific fields could be helpful. For example, I did not include 1712's data in the high goal t-score calculations. This caused averages to change and thus some of the strange cardinal results you see in the final order sort. So, in the example of 708 and 1257: 708 has a higher average and higher standard error than 1257, so, with the low goal specific analysis the higher general average resulting from eliminating teams that aren't competitive low goalers makes standard error more important and thus 1257 does better relative to 708 in the low-goal specific one rather than the boulder volume one.
* at least, overvalued for my purposes in picklisting.