Date: 2009-09-12 04:34 pm (UTC)
From: [identity profile] the-locster.livejournal.com
Heh. An interesting result of the Netflix prize was that it became apparant that all people (well, people who rent films anyway) can be objectively placed on about 60 distinct (and orthogonal for the mathematicians out there) feature axes (dimensions) at the very least. That said the majority of variance in the rating data that could be accounted for was accounted for by the first few features - so charts like this with 3 axes probably make a lot of sense.

As an example, here are the filsm from opposite ends of the first axis from a matrix decomposition of the data. It could probabyl be best described as the 'intelligence axis' (maybe?)

From:
http://sifter.org/~simon/journal/20061027.2.html


Category 1:

Pearl Harbor (2001)
Coyote Ugly (2000)
The Wedding Planner (2001)
Armageddon (1998)
Maid in Manhattan (2002)
On Deadly Ground (1994)
The Fast and the Furious (2001)
Gone in 60 Seconds (2000)
How to Lose a Guy in 10 Days (2003)
Speed 2: Cruise Control (1997)
Jack Frost (1998)
Sweet Home Alabama (2002)
Miss Congeniality (2000)
S.W.A.T. (2003)
Eddie (1996)

Vs.

Lost in Translation (2003)
The Royal Tenenbaums (2001)
Dogville (2004)
Eternal Sunshine of the Spotless Mind (2004)
Punch-Drunk Love (2002)
Before Sunset (2004)
The Life Aquatic with Steve Zissou (2004)
The Mother (2003)
Primer (2004)
Adaptation (2002)
Sideways (2004)
Brothers (2005)
Napoleon Dynamite (2004)
Fahrenheit 9/11 (2004)
Sin City (2005)

Date: 2009-09-14 10:54 pm (UTC)
From: [identity profile] mmcirvin.livejournal.com
Huh, Sin City?

I'd have guessed there would be a "girl/boy" axis, and I'm not sure if Axis 2 is that or not, since all the stuff at one extreme is episodes of "Friends"; it could just as easily be "movies/TV" or "Friends/anything other than Friends".

But, then again, girls and boys have a marked tendency to live together, which would blur the girl/boy axis in a Netflix queue.

Date: 2009-09-14 10:59 pm (UTC)
From: [identity profile] mmcirvin.livejournal.com
...Axis 3, meanwhile, has a prominent "Democratic/Republican" flavor.

Date: 2009-09-15 07:44 pm (UTC)
From: [identity profile] the-locster.livejournal.com
Yeh the Michael Moore and anti-Moore films. The axes are entirely fascinating for sure :)

Date: 2009-09-15 07:41 pm (UTC)
From: [identity profile] the-locster.livejournal.com
Regarding a girl/boy axis, it's an interesting point. It seems that what we consider girly and boyish traits don't exist as a distinct or primary axis/feature. They exist for sure but they are made up from a mixture of the actual primary features which might best be tagged by concepts such as left/right political leaning, libertarian/authoritarian and IQ.

In terms of actual gender there's also a considerably less than 1.0 correlation between being a girl and being 'girly' :)

As for men/women living together and using the same netflix account, I actually attempted to model multiple account users:

Netflix Factorization: Multi-User Accounts

It didn't produce a better model which could mean that the effect is very weak or the approach is flawed. I actually just spotted a problem and possible remedy just now. There's some doubt in my mind as to whether this sort of information is actually in the data to 'eek' out and is sufficiently significant/strong enough to be detected.

Date: 2009-09-12 05:16 pm (UTC)
From: [identity profile] lebeautemps.livejournal.com
Yup. Oh no, wait.

Date: 2009-09-12 09:18 pm (UTC)
zz: (Default)
From: [personal profile] zz
i'm actually considering changing my "do you consider yourself disabled?" answer at work from no to yes, wrt aspergers, as an insurance policy against being fucked over by inflexible bureaucracy.

Date: 2009-09-13 12:21 pm (UTC)
zz: (Default)
From: [personal profile] zz
yeah, i explored the NHS' headpoking services 3-4 years ago re depression & social anxiety, and asked to be tested for it.

Date: 2009-09-12 09:35 pm (UTC)
From: (Anonymous)
On the periphery of yellow and blue.

Greetings from London.

Date: 2009-09-13 09:46 am (UTC)
From: [identity profile] cartesiandaemon.livejournal.com
Ooh, that's very interesting, because that fits more-or-less hit-and-miss my conception of the terms, but I know some people use them interchangeably, or with different connotations.

Date: 2009-09-13 03:43 pm (UTC)
From: [identity profile] scifantasy.livejournal.com
OK, for the record? I love your diagram. Best categorization of the terms I've ever seen.

(I got here from [livejournal.com profile] thette, who got it from [livejournal.com profile] supergee)

September 2025

S M T W T F S
  12 3 4 5 6
7 8 9 10 11 12 13
14 15 16 17 18 19 20
21222324252627
282930    

Most Popular Tags

Style Credit

Expand Cut Tags

No cut tags
Page generated Sep. 20th, 2025 02:45 pm
Powered by Dreamwidth Studios