Histogram of the ratings to see the actual distribution instead of just the average :implemented:

[aej]

I've already suggested this on a comment to another post here ( https://mangadex.org/thread/22932) in passing.
Basically I'm suggesting what mangaupdates already implements on their website:

The image shows a histogram as it is implemented on mangaupdates (i.e. a chart where we can see what proportion of people voted 10, 9, 8, ...). This can show how "polarizing" some series are. Some manga have very zealous followers (who rate the manga as 10) and detractors (who rate the manga as 1). This would come out as having a rating around ~4 to 7. This makes it indistinguishable from manga with only "meh" followers (who rate it as ~7). Having a histogram would be able to distinguish those manga.
I understand this isn't a necessity, but once this site implements all the "low hanging fruit" & urgent features, I hope they consider this.

 

[Le_Dook]

I think a page showing all kinds of stats for each manga would be even better! Some of the current stats are pretty vague so an in-depth view on all would be amazing

 

[Holo]

This is coming.

 

[aej]

@Le_Dook I made another suggestion for a more in-depth stats here: https://mangadex.org/thread/48371. I can't think of other useful/interesting stats to add to the suggestions though, so if you have ideas you could add them there.

 

[jasty]

Ooo it's in and looks great!

 

[aej]

dang it looks great! I didn't think you'd implement it this quickly

 

[DANDAN_THE_DANDAN]

I'm pretty sure the trend in these histograms everywhere is that every series will have a peak around 10-7 (if it's good), decrease steadily until 2. And a big bump at 1. I'm calling this "otaku trend".

 

[Holo]

I will also do the Bayesian average, and then announce these.

 

[Tikibo]

I think it would better to display ratings with 0 votes in the histogram. Currently some series have histograms with values: {10, 8, 5, 2, 1} etc.

 

[zetsuravez]

hi... i noticed the rating(in manga and such) has this Bayesian average and saw this thread...

what is Bayesian average and difference between this and the ordinary "average"?... also what use is Bayesian average that normal average cant give...

 

[ununseti]

@zetsuravez tl;dr Bayesian average tries to give more accurate ratings for manga that have a larger rating uncertainty due to low number of votes

e.g. say in a search you sort by score and at the top is some doujinshi where 2 people rated it, giving it a 10. This score might not be representative of the Mangadex population as a whole, so it's not very accurate. Bayesian average works by shifting ratings with less votes closer towards some prior mean (e.g. site-wide average, or something like that). Lots of sites like AnimeNewsNetwork, MyAnimeList, MangaUpdates, Amazon, etc. already use something like this.

 

[zetsuravez]

@ununseti i see... thanks... kinda hard to understand how the Bayesian average works... but this one looks like it helps manga with low votes counts in terms of searching by votes and such...

 

[sakamoto]

@zetsuravez, kinda late reply, but yeah, in layman's term, Bayesian average:
[ul]reduce extreme values from low number of votes (1 to certain selected number), e.g. a manga with 1 rating of 10 shouldn't sit above a manga with 100 ratings of 9.
it basically bring the rating closer to global average, and only allow it to reach the very top (or bottom) as the number of votes increases
[/ul]
You can read further here:
reference #1: https://stats.stackexchange.com/questions/15979/how-to-find-confidence-intervals-for-ratings/16035
reference #2: http://thebroth.com/blog/118/bayesian-rating.html

mangadex forum:
Some demonstration here 1# https://mangadex.org/thread/22932