Top 100

I don't understand why Camel's SnowGoose is number 13 when half of the albums in front have a lower rating...

Hmm... I don't really understand... The Camel album has TONS of reviews and a rating better than half the ratings in front of it...

How does it work? What's the formula?

Ah, thanks.

But.... why is the number of ratings considered a factor?

I mean, if tomorrow 50.000 people are going to review the worst prog album on the archives:

1.0 * 1.0 * 1.0 * 1.0 * 50.000 = Nr 1.




Edited by LuciferJim

LuciferJim wrote: Ah, thanks. But.... why is the number of ratings considered a factor? I mean, if tomorrow 50.000 people are going to review the worst prog album on the archives: 1.0 * 1.0 * 1.0 * 1.0 * 50.000 = Nr 1.

Your formula isn't correct: 1.0 * 1.0 * 1.0 * 1.0 * log(50.000) = ?

Have a look of how the value of the log() function computes for different input values ... you'll notice the dampening effect. the higher the input value, the less is the effect of a further increase of it.

log(5)   = 0.69

log(10)  = 1

log(20)  = 1.3

log(30)  = 1.47

log(40)  = 1.6

log(50)  = 1.69

log(60)  = 1.77

log(70)  = 1.84

log(80)  = 1.90

log(90)  = 1.95

log(100) = 2

log(110) = 2.04

log(120) = 2.07

log(130) = 2.11

log(140) = 2.14

log(150) = 2.17

...

...

 

log(500) = 2.69

log(5000) = 3.69

log(50000) = 4.69

...

Edited by MikeEnRegalia

^ I know what you mean.

Last week I thought about it and had an interesting idea: What if we only count the positive ratings as number_of_ratings (ratings of 3, 4 or 5 stars)? That would solve the problem IMO ... in your example in the above post the albums might still be "reversed", but if the second album has 140 more "satisfied users", why not prefer it to the other one? There is no way to say which is the "better" album, some people will say that they prefer the higher rated, others will choose the more popular one.

LuciferJim wrote: Have you ever considered to just base the list on the ratings, but only with a minimum of votes, say 30 or so?

no, not really ... well, not for THAT list. But I think that there should be other top 100 lists ... top 100 new albums, top 100 albums with very few reviews, top 100 albums with "some" reviews ...

I'll ask M@x about these things when he has more time on his hands.

I think imdb rating calculation method has it all (minimum votes, ignoring votes given with bad intentions (trying to raise the rating of an album by giving very low marks to others), votes versus ratings, etc.). Have a look at the following explanation given in the top 250 page. It might give an idea.

The formula for calculating the Top Rated 250 Titles gives a true Bayesian estimate:

weighted rank (WR) = (v ÷ (v+m)) × R + (m ÷ (v+m)) × C

 where:
  R = average for the movie (mean) = (Rating)
  v = number of votes for the movie = (votes)
  m = minimum votes required to be listed in the Top 250 (currently 1300)
  C = the mean vote across the whole report (currently 6.8)

for the Top 250, only votes from regular voters are considered.

MikeEnRegalia wrote: ^ lol ... I recently browsed the imdb and examined this formula ... we'd have to apply it to the data in the archives and compare results. The question is: What would be a reasonable value for the minimum number of votes ... 20, 50, 100?

Well, that depends, whether or not you feel these albums should belong to the list:

Camel - Moonmadness (107)
Jethro Tull - Aqualung (102)
Rush - A Farewell to Kings (95)
Gentle Giant - In a Glass House (80)
Gentle Giant - The Power and the Glory (63)
Supertramp - Crime of the Century (60)
Porcupine Tree - The Sky Moves Sideways (55)
Frank Zappa - Hot Rats (55)
Opeth - Still Life (49)
Queen - Queen II (45)
Mahavishnu Orchestra - Birds of Fire (40)
Frank Zappa - We're Only In It For the Money (37)
Focus - Hamburger Concerto (33)
Mr. Bungle - Disco Volante (22)
Frank Zappa - The Grand Wazoo (20)
Fates Warning - Still Life (12)
Pink Floyd - PULSE (8)
Camel - Coming of Age (5)
Queensryche - Operation LIVECRIME (4)


To name a few albums and the number of votes.


Edited by LuciferJim