LogFAQs > #882472617

LurkerFAQs, Active DB, Database 1 ( 03.09.2017-09.16.2017 ), DB2, DB3, DB4, DB5, DB6, DB7, DB8, DB9, DB10, DB11, DB12, Clear
Topic List
Page List: 1
TopicVisual Novel General v3 "The toaster oven is a sacred treasure"
kirbymuncher
07/07/17 1:19:14 AM
#25:


Itachi157 posted...
What's up with the "Bayesian rating" on vndb?

How does that differ from just the normal rating? I tried googling it and the Wikipedia page wasn't really enlightening.


Normal rating is just the average of all votes on the VN. Pretty simple.

Bayesian rating, I don't know all the exact math behind, but the idea is that scores are generally pulled towards the overall site average. VNs with fewer votes are affected significantly more by this. This means a VN with very very few votes will score near the overall average of the site regardless of what the actual votes were (eg https://vndb.org/v16435). VNs with a ton of votes will have their average and bayesian ratings more or less the same.

The reasoning behind this method is basically that you'd expect VNs to, on average, be pretty close to the average VN score. So when something has very few votes, you still score it relatively close to the average because the sample size is low and it might just be a few guys with weird taste that you don't want completely representing the VNs score.

Like imagine some random obscure thing that only one person ever read and they rated 10/10. It would be sort of silly to regard this as the highest rated VN and that's basically the thing bayesian helps solve.


Personally I don't like it since the population of VNDB is mostly english readers, meaning scores on the comparatively less-read untranslated VNs are kind of worthless since they get skewed towards the middle. Then again I don't really look at vndb ratings very much in the first place
---
THIS IS WHAT I HATE A BOUT EVREY WEBSITE!! THERES SO MUCH PEOPLE READING AND POSTING STUIPED STUFF
... Copied to Clipboard!
Topic List
Page List: 1