Band Ratings - Suggestions - Forums : Best Ever Albums

Xavygravy

what are we going to do about them?

They're a joke as they are right now. Most bands have 0-3 ratings.

How can we encourage more users to rate bands?

videoheadcleaner
formerly Harkan

Gender: Male
Age: 38

Average out what they rate their albums to an average band rank?

Xavygravy

harkan wrote:

Average out what they rate their albums to an average band rank?

That's an interesting idea. Perhaps on band rating calculations, instead of drawing from the site average in the Bayesian average equation, it could draw from the average rating of that artist.
e.g. The Beatles album average may be 83 and Nickelback's album average may be 51, so if I give both bands a 100 (which I wouldn't), then The Beatles would be around 86/87 whilst Nickelback would be around 54/55, despite both receiving 100s.

To any user of the site, however, it doesn't look any different or any more complicated (just more accurate).

To encourage rating bands, perhaps instead of a song quote, it would show a little reminder from time to time with all kinds of "did you know?" things, including "Don't forget to rate your favourite bands!" or something like that.

albummaster
Janitor

Gender: Male
Location: Spain
Site Admin

Xavygravy wrote:

To encourage rating bands, perhaps instead of a song quote, it would show a little reminder from time to time with all kinds of "did you know?" things, including "Don't forget to rate your favourite bands!" or something like that.

I think the main issue is that there are not enough ratings, so I think it's a good idea to try and promote more.

Regarding the calculation itself, the best improvement I can think of would be to take into account the individual's average member rating when calculating the overall average.

Code:

The average rating is not a simple arithmetic mean, instead it is calculated using a Bayesian average:
Average Rating = (n ÷ (n + m)) × av + (m ÷ (n + m)) × AV

where:
av = mean average rating an item has currently received.
n = number of ratings an item has currently received.
m = minimum number of ratings required for an item to appear in a 'top-rated' chart (currently 10).
AV = the site mean average rating.

Therefore, I think av in the above formula should be derived from a 'weighted' member rating and not just a mean.

e.g. if someone's average rating is 70 and they rate something a 90, that has a bigger effect than someone rating a 90 who usually rates a 85 (and the same in reverse).

albummaster
Janitor

Gender: Male
Location: Spain
Site Admin

I just want to point out here how the current system works as I think there might be a bit of confusion regarding this at the moment. Let's say an album has already received the following ten ratings:

90/100
100/100
60/100
80/100
90/100
100/100
60/100
80/100
90/100
100/100

The mean average is therefore: 85% (850/1000) and the Bayesian is 82.5%

If I now rate this album 5/100 (well outside of the norm). This now causes the mean to decrease 7.3% to 77.7% (855/1100).

But, remember BEA uses a Bayesian average, so let's plug in the variables and see what happens:

Code:

Average Rating = (n ÷ (n + m)) × av + (m ÷ (n + m)) × AV

Average Rating = (11 ÷ (11 + 10)) × 77.7 + (10 ÷ (11 + 10)) × 80 = 78.8% (a 3.7% decrease)

So, instead of being reduced by 7.3%, what happens is the rating changes by half of that, and the more ratings there are, the less the differential will be (and for items with lots of ratings, it will be barely perceptible).

In the opposite scenario, let's assume I want to big up my favourite by assigning a 100. The mean is increased 1% to 86% (950/1100), but look what happens to the Bayesian:

Code:

Average Rating = (11 ÷ (11 + 10)) × 86 + (10 ÷ (11 + 10)) × 80 = 83% (a very slight 0.5% increase)

In conclusion, rating at the high and low extremes probably doesn't have the impact that people might think because each item's rating is based on the Bayesian average. However, I still think that taking into account each individual's average will soften this effect even more and remove the need for people to rate at the extremes.

GARY

Xavygravy wrote:

what are we going to do about them?

They're a joke as they are right now. Most bands have 0-3 ratings.

How can we encourage more users to rate bands?

Hold on Xavygravy, ssslllloooooowwww your roll a bit please.

Let me get this straight, soz I understands dis.

YOU, of all the people on BEA, are complaining about

The BAND RATINGS Shocked

You who have given an 80 rating or higher to

Lady Ga Ga
Mariah Carey
Jonas Brothers
Justin Timberlake
No Doubt
A-ha
Dixys Midnight Runners

Now, different strokes for different folks and this in itself, though rather odd, I can live with.

But when you consider the fact that you gave a 0 or a 5 rating to these Shocked

Led Zeppelin Evil or Very Mad

Iron Maiden Evil or Very Mad

Pink Floyd Evil or Very Mad

The Who

O.K. right then, there are 4 out of my 6 favorite bands Shame on you

And there are many more that you gave either a 0 or a 5 rating to

AC/DC
The Doors
Metallica
Paul McCartney
Elton john
Talking Heads
Ozzie Osbourne
Steely Dan
Public Enemy
Van Halen
Porcupine Tree
Cream
Nirvana
Belle and Sebastian
U2
Motley Crue
Alice In Chains
Kansas

ALL of which are ON MY CHART Evil or Very Mad

Yet YOU are complaining about the band ratings Shocked

_________________
.
I owe $100,000 and wasted 4 years of my life.

And all I got was this silly hat

.

Xavygravy

I'm not talking about "how" bands are rated, I'm talking about the fact that no one is rating bands.

Xavygravy

albummaster wrote:

Code:

The average rating is not a simple arithmetic mean, instead it is calculated using a Bayesian average:
Average Rating = (n ÷ (n + m)) × av + (m ÷ (n + m)) × AV

where:
av = mean average rating an item has currently received.
n = number of ratings an item has currently received.
m = minimum number of ratings required for an item to appear in a 'top-rated' chart (currently 10).
AV = the site mean average rating.

Therefore, I think av in the above formula should be derived from a 'weighted' member rating and not just a mean.

e.g. if someone's average rating is 70 and they rate something a 90, that has a bigger effect than someone rating a 90 who usually rates a 85 (and the same in reverse).

I think this is a good idea. Please implement it for all types of ratings.

This will also mean my extreme low and high ratings won't make the impact people think, because my 0s will be closer to my low album average rating, meaning they are weighted less.

Also, who's to say how I use the rating scale? There are no guidelines for it. Some people rate harshly, giving only one 100, and the rest in the 50s range. Some only rate in 100s. How is this any better? Everyone has there own 'systems'.

GARY

Xavygravy wrote:

albummaster wrote:

Code:

The average rating is not a simple arithmetic mean, instead it is calculated using a Bayesian average:
Average Rating = (n ÷ (n + m)) × av + (m ÷ (n + m)) × AV

where:
av = mean average rating an item has currently received.
n = number of ratings an item has currently received.
m = minimum number of ratings required for an item to appear in a 'top-rated' chart (currently 10).
AV = the site mean average rating.

Therefore, I think av in the above formula should be derived from a 'weighted' member rating and not just a mean.

e.g. if someone's average rating is 70 and they rate something a 90, that has a bigger effect than someone rating a 90 who usually rates a 85 (and the same in reverse).

I think this is a good idea. Please implement it for all types of ratings.

This will also mean my extreme low and high ratings won't make the impact people think, because my 0s will be closer to my low album average rating, meaning they are weighted less.

Also, who's to say how I use the rating scale? There are no guidelines for it. Some people rate harshly, giving only one 100, and the rest in the 50s range. Some only rate in 100s. How is this any better? Everyone has there own 'systems'.

Personally, I have no problem with you giving 0 and 5 ratings. That is up to you of course, besides, I am a low rater myself and use the full 0-100 range also.

But you pretty much rated the bands that have made about 1/2 of the albums on my chart a 5 or less. Sad

I had to say something Anxious

_________________
.
I owe $100,000 and wasted 4 years of my life.

And all I got was this silly hat

.

Xavygravy

GARY wrote:

Xavygravy wrote:

albummaster wrote:

Code:

The average rating is not a simple arithmetic mean, instead it is calculated using a Bayesian average:
Average Rating = (n ÷ (n + m)) × av + (m ÷ (n + m)) × AV

where:
av = mean average rating an item has currently received.
n = number of ratings an item has currently received.
m = minimum number of ratings required for an item to appear in a 'top-rated' chart (currently 10).
AV = the site mean average rating.

Therefore, I think av in the above formula should be derived from a 'weighted' member rating and not just a mean.

e.g. if someone's average rating is 70 and they rate something a 90, that has a bigger effect than someone rating a 90 who usually rates a 85 (and the same in reverse).

I think this is a good idea. Please implement it for all types of ratings.

This will also mean my extreme low and high ratings won't make the impact people think, because my 0s will be closer to my low album average rating, meaning they are weighted less.

Also, who's to say how I use the rating scale? There are no guidelines for it. Some people rate harshly, giving only one 100, and the rest in the 50s range. Some only rate in 100s. How is this any better? Everyone has there own 'systems'.

Personally, I have no problem with you giving 0 and 5 ratings. That is up to you of course, besides, I am a low rater myself and use the full 0-100 range also.

But you pretty much rated the bands that have made about 1/2 of the albums on my chart a 5 or less. Sad

I had to say something Anxious

another reason I use extreme scores is to balance out other users... who also use extreme scores.
e.g. Lady Gaga is rated by many really low, so I rate her 100 to balance this out, even though in reality she's only worth around a 70 or so. Those ones you mentioned, that are on your chart, were probably rated 100 so many times that I felt they were a little overrated in comparison to others. In most cases I don't actually hate those bands.