shoop (mcfnord) wrote in lj_nifty,

LJ Metro Concentrations

I've collected about 57,000 cities from a city field found in LJ user foaf feeds. This field has its problems. It might be blank 90% of the time, people write mumbo jumbo in it, and people often write the same city's name more than one way. Still, it can sense large city names.

These are the most common cities among actively-posting LJers's:


users | city
926 | Moscow
857 | Москва (Moscow)
346 | Seattle
205 | San+Francisco
202 | Toronto
200 | Chicago
197 | London
195 | Portland
159 | New+York
151 | Austin
145 | Los+Angeles
138 | Melbourne
135 | Санкт-Петербург (St. Petersburg)
128 | Boston
123 | Minneapolis
107 | Brooklyn
103 | Atlanta
103 | Vancouver
  • Post a new comment

    Error

    Anonymous comments are disabled in this journal

    default userpic

    Your reply will be screened

    Your IP address will be recorded 

  • 17 comments

lil_m_moses

January 20 2009, 14:47:17 UTC 5 years ago

What kind of timeframe does that look at? In other words, would the distribution shift if you were to look again about 8 hours from now?

mcfnord

January 20 2009, 23:30:46 UTC 5 years ago

It's more like a timeframe of 6 months. Yesterday I did 9,463 such samples, of 611,357 total samples. Perhaps that's a clue, perhaps not.

borgseawolf

January 20 2009, 14:50:26 UTC 5 years ago

So basically if I'm looking for people outside USA and Russia... I should look elsewhere? :P

da_lj

January 20 2009, 15:08:25 UTC 5 years ago

Take another look at the top four cities. One of them isn't in the US or Russia. :P

1783 | Moscow
346  | Seattle
205  | San+Francisco
202  | Toronto


For that matter, London plus Melbourne together would place at #3.

I'm curious what time-frame the foaf feeds call "active".

klgrem

January 20 2009, 16:48:53 UTC 5 years ago

Moscow is not in Russia?
Where did San Francisco go if it's not in the US?

bitter_crimson

January 20 2009, 17:24:55 UTC 5 years ago

Toronto is in Canada (not the U.S. or Russia), is what the above commenter was saying.

klgrem

January 20 2009, 17:35:45 UTC 5 years ago

Right. I misread his reply. Twice haha. I guess I need new bifocals. Thanks for pointing that out. Though I knew that, this time I read their reply right. ;)

scearley

January 20 2009, 17:38:19 UTC 5 years ago

I'd say Seattle is closer to being Canadian than Toronto is.

sinnick

January 20 2009, 18:13:39 UTC 5 years ago

You're from Vancouver originally, aren't you?

glowing_dragon

January 20 2009, 20:19:55 UTC 5 years ago

VANCOUVER REPRESENT!

mcfnord

January 20 2009, 22:06:56 UTC 5 years ago

Ha ha sly Canada!

You know, I have heard that Singapore is a major country for LJ... and I have seen clues of this... I was surprised when I couldn't find any cities there. Hmm.

I call active... it's a pragmatic definition but it should make sense... I sample LJ post count as frequently as every 60 days. So it's probably true that some time in the last 3 months or so, two samples at least 60 days apart detected the same post count. I do not trigger a review because I don't sense how old the old samples are... but I would say that contributes to accuracy, at the cost of some currency. I do walk a large LJ map pretty fast, however, and consider pulling a foaf sample (if the old one is older than 60 days) whenever i pull an fdata sample. all of this is limited by the fact that i really just started tracking data 3 months ago.

There are other problems with this (and some people's post count is lower, if they went on a post-killing spree), but it is a pretty good sensor of an inactive LJ, do you think?

hammond

January 20 2009, 22:59:27 UTC 5 years ago

Why would you put London and Melbourne together though?

psychokitten76

February 3 2009, 03:47:48 UTC 5 years ago

Because there is a Melbourne in England too.

lushlush

January 20 2009, 15:35:45 UTC 5 years ago

What, no Manila?

pne

January 20 2009, 16:37:02 UTC 5 years ago

Perhaps that's obscured by people listing "Metro Manila" as well and therefore splintering the numbers -- maybe even "Maynila".

mcfnord

January 20 2009, 23:34:15 UTC 5 years ago

There's a big hairy normalization problem in the data.

Cebu+City/Manila
Las+Pi%C3%B1as,+Metro+Manila
Manila
Manila+Below
Manila/Cebu
Manila,+Philippines/New+York+City
MetroManila
Metro+Manila
Metro+Manila,+Philippines
Metropolitan+Manila
New+Manila
Quezon+City,+Manila
San+Juan,+Metro+Manila
Tondo,+Manila
Manila
Manila+Below
Manila/Cebu
Manila,+Philippines/New+York+City
Cebu+City/Manila
Las+Pi%C3%B1as,+Metro+Manila
Manila
MetroManila
Metro+Manila
Metropolitan+Manila
New+Manila
Quezon+City,+Manila
San+Juan,+Metro+Manila
Tondo,+Manila

cdvla313

February 21 2009, 06:02:12 UTC 5 years ago

And if you combined New York and Brooklyn, it would be number four. Since you know, Brooklyn is part of New York City, not a separate city.