What happened to TOP 100???

^ not true. In fact you could say that the new algorithm is all about the number of ratings, since they are used as weights. In essence the system computes the average of all ratings, and the average *number* of ratings for an album. Let's say that the average number of ratings per album is 100, and the average rating of all albums is 3.5. The weighted average is then computed as:

(100*3.5 + n*avg)/100*n

where

n = number of ratings for the album
avg = average rating of the album

now if avg=5 and n=1 we get:

(100*3.5 + 5)/101 = 3.51

which means that because of the low number of ratings the album is virtually glued to the total average. However, if n=100 we get

(100*3.5+100*5)/200 = 4.25

and if n=200 we get

(100*3.5+200*5)/300 = 4.5

see? I'm not saying that this is an ideal thing ... but it *does* depend on the number of ratings.

chopper wrote: sellingengland wrote: ahhhhh, excellent to see Hamburger Concerto up at 10 there. What is with peoples obsession of Close to the edge? Fargile is a far better album! Peace out Therein lies one of the main problems with charts like this. You are mistaking your opinion for fact. You  may prefer "Fargile" (sic) to CTTE but the fact that CTTE is generally higher in the chart means that you are in the minority. It's all subjective.

Yeah I undertand that, I was just making the point that people on this site were giving higher ratings to CTTE than fragile, and that i disagreed with this. Cheers for the correction by the way .

Peace out

Joolz wrote: Seyo wrote: Eetu Pellonpää wrote: chopper wrote: I still believe that proto-prog and prog-related acts should not be included.   I'll second this!   Me too, remember, these are not "real" prog rock! If this is so, then how can they be getting high enough marks to be in the top 100 in the first place?

sellingengland wrote: ahhhhh, excellent to see Hamburger Concerto up at 10 there. What is with peoples obsession of Close to the edge? Fargile is a far better album! Peace out

ahhhhh, excellent to see Hamburger Concerto up at 10 there. What is with peoples obsession of Close to the edge? Fragile is a far better album!

Peace out


Edited by sellingengland - July 12 2007 at 11:11

Eetu Pellonpää wrote: chopper wrote: I still believe that proto-prog and prog-related acts should not be included.   I'll second this!

I am very sceptical. Some results are just plain weird. I am a huge PT fan, but the DVD Arriving Somewhere... in 12th place? Sorry, but I think In Absentia is a far better release. To me it seems this list does not take enough into account the fact that scores tend to become lower with more votes. 4,5 with 100 and 300 votes is far from the same. The 4,5 album with 300 votes would probably have something like 4,3 in score if it had 300 votes. People most enthusiastic about an album tend to vote on it first, like fans. I think that should be the main concern when making an algorithm. The new one seems very complex, but I think it fails to generate reliable results. I think it should be worked more on. Perhaps you can give us different algorithm proposals to vote over? I think it’s a good idea to let collaborators votes count more, at least.

After reading more about this new rating system, it seems like the number of votes are not taken into account, other than the fact you need a certain number of votes to get into the list. That is just plain stupid and a big mistake. As I already have mentioned, more votes leads to lower score, it is just the law of scoring, and one cannot ignore that. It seems like most fail to realize this.

Edited by Sofagrisen - July 12 2007 at 09:28

darqdean wrote: Sckxyss wrote: M@X wrote: Yes , there is been a upgrade to the algo , it's now based on the more adequate weighted average calculation (more info here : http://en.wikipedia.org/wiki/Weighted_average )   ;-) Any comments sor far ? What exactly is it weighted on? And, I don't see how an album with one rating of 4 stars could get an average rating of 3.46. Could you explain more? I'm curious about exactly how this is calculated. I don't care either way about the chart positions (obviously I like to see Floyd do well, but that's personal bias), however as a technical-type person I am intrigued by the maths behind the algorithm.   I haven't a clue how the weightings are done but I have a feeling that the weighting is based upon the ratings of other albums by the same artist.   Take a look at this: QUATERMASS "The Long Road"  a single 190-word review by a non-colaborator/reviewer giving it a 1-star rating, yet the total is 3.51.   The only way I can see this occuring is if the 3.66 rating for Quatermass's previous album is taken into concideration.   This would explain CTTE sudden drop from the top three.

 The title also explains why I don't actually understand the algorithm!

MikeEnRegalia wrote: Easy Livin wrote: The number of reviews and the ratings are the only factors taken into account for the algorithm.   The weighting simply give more emphasis to reviews by Prog reviewers and Collaborators, and to ratings which include a written review. This was done in response to the general distaste for ratings without reviews. I'm curious ... how does the number of reviews affect the ranking? It seems like it was removed completely ... the average is the only thing that counts now.