Re: What's the point of using Bayesian average for album rat

albummaster — Thu, 27 Jun 2019 05:16:08 GMT

Author: albummaster

Posted: 06/27/2019 09:16

There's drawbacks with all types of average (a mean is especially prone to gaming), so the choice to use a Bayesian average factors in these types of considerations. Yes, happy to add the option of viewing by mean or Bayesian average. The 'top rated' albums page is probably the best place for this (it already provides the ability to adjust the number of ratings, so it should be a relatively small change to allow you to view by other types of average).

What's the point of using Bayesian average for album rating?

jpajovich — Wed, 26 Jun 2019 12:27:50 GMT

Author: jpajovich

Posted: 06/26/2019 16:27

I get the point of using it for tracks since that is the only way they are organized, but the overall chart is organized by rank score which is only determined by chart position.

I understand the point of using a Bayesian average lowers the effect of outliers and does not allow albums with only a few ratings from being rated too high. On the other hand, it does exactly that. The use of the Bayesian average rewards the popularity more than the actual average rating (which is what the rank score does essentially as well). Instead of its intended goals it just appears more as a skew of the true mean.

Possibly could there be a way to see both the true average as well as the Bayesian? With the option to display the chart using each (maybe even putting a minimum number of ratings to display the chart as well)?

Best Ever Albums

Re: What's the point of using Bayesian average for album rat

What's the point of using Bayesian average for album rating?