[ 3 / biz / cgl / ck / diy / fa / ic / jp / lit / sci / vr / vt ] [ index / top / reports ] [ become a patron ] [ status ]
2023-11: Warosu is now out of extended maintenance.

/sci/ - Science & Math

Search:


View post   

>> No.9675049 [View]
File: 335 KB, 1250x1072, Screen Shot 2018-04-15 at 10.03.25 PM.png [View same] [iqdb] [saucenao] [google]
9675049

This has been on my mind a lot recently.

How do you accurately rank based on scores? For example, in the picture (which is a screenshot of the website MyAnimeList), they have a way of ranking things that makes a lot more sense than in IMDb.
For example, if you look at "Top Animations" in IMDb, you consistently see shows/movies that have very few votes out-ranking more popular films.

With the MAL list, though, the #4 show has a slightly higher score than the #5 show, despite the 5th show having many more members. This system is apparently slightly biased towards less people voting (but I may be wrong here)

I guess my question is, what would a good formula be for ranking shows?
Something I came up with quickly, and works decently for sufficiently large numbers of votes, is:
[math]
Score = (Avg) \cdot [ log_{10}(V) + 1]
[/math]
Where V is the number of people that have rated a show, and "Avg" is the average of the scores.

One small problem is that, with this, it's possible for a show with an 8.5 average to be ranked higher than a show with a 9.0 (although in this case, if the show with an 8.5 had [math] b [/math] people voting, and the show with a 9.0 had [math] a [/math] people, then [math] b \geq a^{1.05} [/math] for this to happen)

Is this a good scheme? How can I improve it?

Navigation
View posts[+24][+48][+96]