Whi****ll request

[Snurfen]

Just saw this (http://www.tek-tips.com/geography.cfm) on another forum, I would find it interesting if someone with the correct access could perform a query and post similar geographical distribution for us at CTH.

Not productive I know, but interesting none the less. :cool:

[Mr Bean]

I know Dai has a stats box over at his place Snurf. At the bottom of the main page.

I look at it from time to time as I find things like that interesting.

Well, maybe not interesting but you know what I mean.

[Snurfen]

how on earth did wh1****ll fall foul of the * monster? Odd.

Yes, my floating Beanmiester, a sad admission, but I too have a wry interest in this sort of thing. Adds a little quality of life, n'est pas?

*EDIT* w-h-i-s-t-f-u-l-l

[dammit]

Best I can do mate.
http://img.photobucket.com/albums/v194/gedchris/f88dfb4c.jpg

[PurestLight]

Very interesting :confused: :confused: except I dont have a clue what you lot are talking about :rotflmao:

Does 'wi****l' 'have the same *** effect. I guess Ill see when i submit my completely unnecessary reply

http://bestsmileys.com/signs15/18.gif

*edit* its the **** part of it lolol (s.t.f.u.)

[Snurfen]

have a look at the link in the first pos, me-julie!

[PurestLight]

oohhhhh I see. (I did look already but I looked again and concentrated) You want to know where everyone comes from, and in what %. :)
It looks like a MiShY kind of a job to me :D

[black mirror]

:confused: :confused: :confused: :confused: :confused:

[oracle128]

I'd do it, but I'd need the full site databases, and that aint gonna happen. Mishy is the only one with access to that. You'd also need a way to automate the IP/Geographic lookup, but all the services I've seen that do that cost money.

[Snurfen]

My original thought, Ora, was something along the lines of scanning the tables holding the membership details and pull out the location fields.
On reflection, this wouldn't be accurate as (a) not many members give location details and (b) they aren't necessarily in a format that could be easily interpreted for output to a meaningful list.

Ah well, neither my original nor your recent suggestion are easy/low cost options.

[Nick Grana]

Mishy may already have these stats but not posted publicly.
He'll probably chime in with a quick wit.:cool:

[MishY]

GeoIP is good but still not as accurate as some would like. I can give u a guestimate however :p

90% USA
4% UK
3% AU
1% NZ
2% lots of others

[Snurfen]

LOL! thanks guvnor :thumbsup:

[dammit]

4% UK? That's not good :hmm:

[Rainbow32]

4% UK? That's not good
Yeah that's only 2% more than lots of others.

[oracle128]

My original thought, Ora, was something along the lines of scanning the tables holding the membership details and pull out the location fields.
On reflection, this wouldn't be accurate as (a) not many members give location details and (b) they aren't necessarily in a format that could be easily interpreted for output to a meaningful list.Yeah I already considered that, but it's not feasible for the reasons you stated.

Really, if there were any stats I'd be interested in, it would be stuff like:

(per user)
Age vs OS
OS vs Location
Average "Average Post Per Day"
Post Count vs Topics Started
Age vs Post Count
Post Count vs PM/Email Count (are PM/Email counts even stored?)
Tech Posts (the OSs, Hardware's, Software's (minus SD&R), AE) vs Non-Tech Posts (OD, Jokes, C&S, SD&R)
OS vs Posts In That OS's Category
Posts In OS Section Of User vsPost Not In That OS Section
Topics Started in Tech Categories vs Starsign (maybe Saggitarians are particularly unlucky?)
Topics Started in Tech Catgeories vs Chinese Zodiac Animal

single figure
# of 'Active' Members (those who have minimum x posts, and minumum y posts per day rate)
# of Tech Topics Resolved (difficult because it needs to be determined manually, and there may be ambiguities over whether the topic was solved or not)

[AnnMarie]

LOL, you forgot race and gender (optional of course) :rotflmao:

Geez Oracle, you have missed your calling. The Census guys would love you. :D

[black mirror]

You also forgot Oracle128 "what colour socks ??":D

[chryssi2001]

Yeah that's only 2% more than lots of others.

Well i am in the 2%, but that's interesting... Do those % include members and subscribers?
If not can we have separate ones please?

[AnnMarie]

Hi chryssi, smack me with a daffodill and call me a curiosity queen but why are you interested in the breakdown between members and subscribers in the 2% category? :D

[chryssi2001]

Hi chryssi, smack me with a daffodill and call me a curiosity queen but why are you interested in the breakdown between members and subscribers in the 2% category? :D

:rotflmao: Hi AnnMarie i won't smack you don't worry, http://www.fadzter.com/smilies/flower.gif i am not asking especially for 2% i was asking a separation of % totals to member and subscribers. The 2% is a minority and i expected that!!! BTW as for curiosity queen title http://www.world-of-smilies.com/html/images/smilies/engel/carlydevil.gif (http://www.world-of-smilies.com) can we share it please?;)

[uripyores]

Someone should start up one of these maps and we could all stick our pins in it.


Create your own map. (http://www.frappr.com/c/user/createamap)

[black mirror]

Someone should start up one of these maps and we could all stick our pins in it.


Create your own map. (http://www.frappr.com/c/user/createamap)


Ask Lufbra.. he has one in daves pub perhaps he would lend it to us:D

[chryssi2001]

Someone should start up one of these maps and we could all stick our pins in it.


Create your own map. (http://www.frappr.com/c/user/createamap)

That's a nice idea!! ;)

[zipulrich]

Are the Canadians in the 'Other' category?

[chryssi2001]

Are the Canadians in the 'Other' category?

http://www.fadzter.com/smilies/wave.gif Welcome to 2%....

[oracle128]

Actually, all the stats I'm after would either be directly beneficial for CTH, or beneficial from a a more general demographics viewpoint (except of course for the two starsign ones, which only have novelty value).

For example, things like Solution Rate (# Solved Tech Threads / # Total Tech Threads) would not only be interesting, but an excellent metric we can use to determine the usefulness of the site; as well as, from the business perspective, being something we could advertise (eg. "40% of all topics are resolved with 24 hours, 20% within 48 hours, 35% are solved eventually, 5% remain unresolved").

[chryssi2001]

Actually, all the stats I'm after would either be directly beneficial for CTH, or beneficial from a a more general demographics viewpoint (except of course for the two starsign ones, which only have novelty value).

For example, things like Solution Rate (# Solved Tech Threads / # Total Tech Threads) would not only be interesting, but an excellent metric we can use to determine the usefulness of the site; as well as, from the business perspective, being something we could advertise (eg. "40% of all topics are resolved with 24 hours, 20% within 48 hours, 35% are solved eventually, 5% remain unresolved").

Nice suggestion Oracle! As i see it...
This site has members from all around the world, so due to different hours most of the times enough posts, delay, in getting answers or resolved, not on purpose, and that would alter the real results of solution rate %!! My poor opinion!! :happy: But on the other hand i can't see another way for this to be done.. and give a true picture of solved threads.

[Snurfen]

I like that idea Ora, a positive way of promoting the site :thumbsup:

[uripyores]

How can you tell if the problem has been solved? The person who had the problem doesn't always let anyone know; I think less than half do so in fact. http://www.smileypad.com/v224/Feelings/Dunno2.gif

[dammit]

How can you tell if the problem has been solved? The person who had the problem doesn't always let anyone know; I think less than half do so in fact. http://www.smileypad.com/v224/Feelings/Dunno2.gif

How true is that.?

[degsy]

You then need to define what is a post that can be classed as resolved?

You would have to disregard posts from the OD, Jokes, Software Review forums.
You would have to disregard Stickies.
You would have to disregard posts that are purely for information such as posts to software or windows updates.
You would have to decide whether a post was a question that could have a definite answer rather than several opinions that would suffice as an acceptable answer.

[oracle128]

You would have to disregard posts from the OD, Jokes, Software Review forums.
You would have to disregard Stickies.
You would have to disregard posts that are purely for information such as posts to software or windows updates.
You would have to decide whether a post was a question that could have a definite answer rather than several opinions that would suffice as an acceptable answer.
The specification of 'Tech Threads' should imply the first two considerations, though to be more specific you could say 'Tech Threads Requesting Help' (not that there's a great many threads in the tech categories that are otherwise'). Because of the nature of gathering this data (ie. having to be done by a human), you could easily only include the relevant threads. To answer your concern uripyores, you'd also obviously have to define a set of criteria to determine if a problem is "fixed", since we don't have a system such as that on Experts Exchange - like I said, it's "difficult because it needs to be determined manually, and there may be ambiguities over whether the topic was solved or not". Confirmation of the working solution is obviously first priority, followed by threads that could only possibly be answered correctly (eg. "How do I take ownership of files?", or "How do I change the resolution?", it would be up to the reviewer to determine, objectively, which ones would fit this description). Threads with a solution, but no obvious confirmation of a result would be discarded from the data collection, as would threads where questions are asked by the helper which don't get answered (eg. probing for further info or clarification); whilst those without a proper posted solution at all would count as a fail.

Obviously it's a huge workload, so you'd probably want to a) have several people do it, and b) take random samples of threads instead of doing them all (taking care not to double-up on threads for the reviewers). Having an impartial 3rd party classify the threads would be best, though it could probably just as easily be a group of members who have no conflict of interest with the site.

Of course, you could simplify the whole activity by looking for 'Problem Tech Threads With A Valid Answer' as opposed to 'Resolved Problem Tech Threads', as it's much easier to identify if a thread has an answer, and to determine whether or not that answer is valid, than it is to determine which threads are resolved.

[degsy]

Obviously it's a huge workload, so you'd probably want to a) have several people do it, and b) take random samples of threads instead of doing them all (taking care not to double-up on threads for the reviewers). Having an impartial 3rd party classify the threads would be best, though it could probably just as easily be a group of members who have no conflict of interest with the site.

There are currently 657,015 posts. Knock yourself out.

[black mirror]

There are currently 657,015 posts. Knock yourself out.

Looks like Oracle128 will be quite busy for a while.. see you next year.. perhaps:D

[oracle128]

That's what random sampling is for; in real life statistics, you hardly ever consider every single data node, the obvious exception being a census or government voting system (in countries where voting is compulsory). You'd take, say, a sample of 100 threads and review them on the relevant status of them. Have 10 people do that, you've then got a more reliable sample of 1000. How long will that take? I estimate about 20 seconds, maximum, for each thread (that includes page load times) if you're only looking for 'valid' answers. Probably about double that time if you need the reviewer to look for 'resolved' threads (considering decision-making time).

So, to find answered threads, 20s * 100 = just over 33 minutes. Searching for resolved threads, you're looking at about an hour and seven minutes (40s * 100).

So yes, it's definitely feasible.