| What is the True Bayesian Estimate?
|
|---|
This is a formula to calculate the average vote, taking into account
the number of people who voted.
|
| Why do you use it?
|
|---|
If one album is rated average 9.5 by a hundred people, and another album is rated 10 by
only one person, which album is better? A top list should take into account the majority
of the votes. So by using the true Bayesian estimate, we make sure the top lists aren't
dominated by albums that are only appreciated by one person.
|
| How does it work?
|
|---|
Every album is assumed to have an average score, and the more people give a high or low vote
for it, the higher (or lower) its score will get. So an album that is voted 10 by one person
will have only a slightly higher than average score, if it's voted 10 by ten persons it will
have a better score, and if even more people vote it a 10 its bayesian score will go closer
to the 10.
|
| So how do you calculate it exactly?
|
|---|
(number of votes * average vote) + (average number of votes * global average of votes)
number of votes + average number of votes
| with the global average of votes 8.5 and average number of votes 50
|
|
| I read somewhere on wikipedia that this estimate is all wrong!
|
|---|
| Fuck them.
|
