Can Anyone Explain me ?

  • 2
  • Question
  • Updated 1 year ago
  • Answered
Can anyone explain me what is the stat "ex: -33 or +55 ". When i open another user account appears her stats (if has public ) and when see "Compared to you" i tried to understand what means, can anyone explain to me what is the formula of this stat?
Photo of SeventhAr7

SeventhAr7

  • 905 Posts
  • 1403 Reply Likes

Posted 1 year ago

  • 2
Photo of Dan Dassow

Dan Dassow, Champion

  • 12763 Posts
  • 12881 Reply Likes
Hi José,

It's easier to use a real example to explain this. Col Needham (IMDb Founder and CEO) made his profile public (http://www.imdb.com/user/ur1000000/).

When I look at his profile I see:


The +40 is a measure of how closely Col Needham's and my ratings match. The technical name for this is Pearson correlation. The value can range from -100 (two people perfectly disagree) and +100 (two people perfectly agree). Most of the values I've seen on IMDb range from around -20 to +60.

If you look at Col Needham's profile you will see something different.

In case your are interested my profile page is http://www.imdb.com/user/ur0127678/ .

See also:
https://help.imdb.com/article/imdb/discover-watch/how-does-compared-to-you-feature-work/G9N7L4DAPLHD...#
Photo of SeventhAr7

SeventhAr7

  • 905 Posts
  • 1403 Reply Likes
Thank you for the explanation
Photo of bderoes

bderoes, Champion

  • 995 Posts
  • 1338 Reply Likes
I have often wondered that, and just never posted the question. I'm surprised there isn't a link right there to an explanation.

For example, on this page:
http://www.imdb.com/user/ur32466103/
I see these stats:

What does the +30 mean?

On this page:
http://www.imdb.com/user/ur54354614/
There is no comparable stat, but this user has only logged ratings of 10.
Photo of Dan Dassow

Dan Dassow, Champion

  • 12763 Posts
  • 12881 Reply Likes
@bderoes,

You posted just after I did and probably did not see my explanation.
Photo of bderoes

bderoes, Champion

  • 995 Posts
  • 1338 Reply Likes
You're right that I wasn't notified of your post. Thanks for alerting me.

The fact that the standard deviation of each set (the subset of my ratings and the subset of the other's ratings) is in the denominator of the Pearson stat explains why no stat is offered on the page where the user rated everything the same (deviation will be 0).

The help text is flawed. It says:
If you set your ratings to public, and visit someone else's activity page that also has public rating, you can see an interesting chart on how their tastes compare to yours. Here's our founder Col Needham's activity page. The page will do a Pearson Correlation between two users ratings and display a statistic about how much your tastes overlap. Note that this feature is under development and the number could change dramatically and you should not make potentially life-altering decisions based on this rather arbitrary number. It will also display a set of films for which your ratings diverged form each most dramatically.
I did a strikethrough on the 2 problems:
  • users should be users'
  • "form each" should probably just be deleted
And since it provides a link to the Wikipedia page, it should probably say that what is displayed is 100 times the Pearson coefficient, and that the calculation is made on the subset of films that both users have rated?

I would post these to the Help Hub thread, but since they only touched some of what I reported on 1 Jan about the Checkins help (actual misinformation is still on the page):
https://getsatisfaction.com/imdb/topics/new-imdb-help-hub-6fxihx3ry4fgd?topic-reply-list[settings][f...
I'll just shrug and move on.

This conversation is no longer open for comments or replies.