What should be the rating calculation formula?

Question

Analytics

On a site dedicated to cars, I want to make the most reliable and unwindable rating of models.
At the same time, I would like to make a rating according to several parameters.
Are there any alternatives to the IMDB formula (borrowed by Kinopoisk - at the bottom of the page )?
Is it possible to simplify it without significant loss of rating reliability? For example, use only the arithmetic mean of votes and their total number.

Answer 1

What didn't you like about the formula as it is? This is the Bayesian probability formula, which works pretty well.
I didn’t really understand what “arithmetic mean of votes and their number” means. The arithmetic mean is the sum of the votes divided by the number of votes. In this case, you will not have an adequate comparison of models with a large number and a small one.
For example, a model with 2000 votes, of which 1999 votes are “5” and one vote is “2”, will receive less than 5 points and will be ranked lower than a model that was voted only once and put “five”. The Bayesian probability formula is just what you need to avoid this situation. See also this article .

Answer 2

Actually, recently there was a translation of a good article on ratings and sorting.
For more details, see the book Building Web Reputation Systems and its blog

Answer 3

I want to ask. Found a multi-parameter rating solution?