Talk:Suggestion/@comment-25490263-20150224220619/@comment-25534439-20150228020611

Well, considering the changes coming, the smallest will be the new Torpedo Cruisers table, with 3 rows.

The core problem with a standard is precisely because the size and arrangement of the data varies significantly. Saying green gets best 1-3, yellow is next 2-5 is silly on small pages. Further, I would argue that even Green is top 10%, Yellow is next 20% doesn't really accurately address the data across the pages.

A proper standardization approach has to address both the presence of outliers (Yuudachi, Ayanami) whose presence is clearly exceptional, as well as how skewed the data is.

For an example of skewed data, look at light cruiser armor: Exempting Yuubari and Katori's exceptionally low armor, there is a group with 59 armor, the Kuma pre-CLTs with 63-65, then a few 68s and a ton of 69s. The norm is the 'high armor' group of 69s which make up almost half the entries. In this case none of them are even outstanding. A grouping this skewed to spread toward the low entries has no notable high entries. (As we have no demarcation for exceptionally low entries, they're ignored. Though, perhaps we should have one? Another time, perhaps)

In my attempt to standardize it I left off with histograms and approximations of skewed normal distributions. Considering the minor benefit it might have relative to the amount of work it would take to develop histograms of all the data, I abandoned it to the current system.

Overall, the point of it is to try to provide an at-a-glance appraisal for which ships are 'the best'. I do wish there was an easy, standardized way, but the current way of guessing is easy and gets it essentially correct.