David Gerard wrote:
Or popular with the readers, at least.
http://tools.wikimedia.de/~leon/stats/wikicharts/index.php?ns=articles&l...
Interesting to-do list task: see what subjects or subject areas are in demand from that list, and make a list of related subjects in the area that aren't covered or could do with serious work.
- d.
Interesting. I am surprised that "Irukandji jellyfish" came so high up on the list... is there some anomoly here?
Not all that surprised about how high Pokemon came up, it is a large fictional universe and I would guess that that universe's main article would get a high result.
Aww... my userpage and user-talk page aren't even on the top lists in their respective categories. :-/
Sjakkalle
On 29/08/06, Sigvat Stensholt sigvats@mi.uib.no wrote:
David Gerard wrote:
Or popular with the readers, at least.
http://tools.wikimedia.de/~leon/stats/wikicharts/index.php?ns=articles&l...
Interesting to-do list task: see what subjects or subject areas are in demand from that list, and make a list of related subjects in the area that aren't covered or could do with serious work.
- d.
Interesting. I am surprised that "Irukandji jellyfish" came so high up on the list... is there some anomoly here?
Jellyfish wasn't on the list at all last night; the statistics are sampling (I think) one in five thousand pageviews, so it's quite likely a glitch. They've only been running for a couple of days, and I'd strongly advise against treating them as reliable until at least a fortnight in.
That said... JonBenet Ramsay is interesting - it surged last night and with no link from the front page. Interesting to see how quickly our use (apparently) reacts to news stories. The sex-related ones - lists of porn stars etc - were consistently high and are now dropping; I wonder where they'll stabilise at.
(It could just be they're more popular at weekends...)
On 8/29/06, Andrew Gray shimgray@gmail.com wrote:
Jellyfish wasn't on the list at all last night; the statistics are sampling (I think) one in five thousand pageviews, so it's quite likely a glitch. They've only been running for a couple of days, and I'd strongly advise against treating them as reliable until at least a fortnight in.
The sampling period is, from my observation, constantly being decreased. It was 5000, then 1000, then 500, now 333. Well, I'm guessing just looking at what numbers have turned up at different times. I don't know exactly how the "+-111%" is calculated, but I would say that we're getting to the stage where the values are somewhat meaningful. With the exception of Jeff Hardy, all the other top 10 entries seem very plausible and topical.
However, I can't see why Irukandji jelly fish would be so high. It's not linked from anything, it's not in the news, and it's not even jellyfish season in Queensland. It could just be a statistical anomaly, or perhaps someone is deliberately messing with the stats?
That said... JonBenet Ramsay is interesting - it surged last night and with no link from the front page. Interesting to see how quickly our use (apparently) reacts to news stories. The sex-related ones - lists of porn stars etc - were consistently high and are now dropping; I wonder where they'll stabilise at.
Yes, that trend is interesting - why did they all show up so high then disappear. Was it a time of day thing? More stats required!! :)
I really hope we will end up with "last 24 hours" and "last 7 days" stats.
Steve
On 8/29/06, Steve Bennett stevagewp@gmail.com wrote:
However, I can't see why Irukandji jelly fish would be so high. It's not linked from anything, it's not in the news, and it's not even jellyfish season in Queensland. It could just be a statistical anomaly, or perhaps someone is deliberately messing with the stats?
Might be on various school national caricular.
On 8/29/06, geni geniice@gmail.com wrote:
On 8/29/06, Steve Bennett stevagewp@gmail.com wrote:
However, I can't see why Irukandji jelly fish would be so high. It's not linked from anything, it's not in the news, and it's not even jellyfish season in Queensland. It could just be a statistical anomaly, or perhaps someone is deliberately messing with the stats?
Might be on various school national caricular.
I'm sure it's place is cemented now that everyone who reads the mailing list has clicked on it ;)
On 29/08/06, Steve Bennett stevagewp@gmail.com wrote:
On 8/29/06, Andrew Gray shimgray@gmail.com wrote:
Jellyfish wasn't on the list at all last night; the statistics are sampling (I think) one in five thousand pageviews, so it's quite likely a glitch. They've only been running for a couple of days, and I'd strongly advise against treating them as reliable until at least a fortnight in.
The sampling period is, from my observation, constantly being decreased. It was 5000, then 1000, then 500, now 333. Well, I'm guessing just looking at what numbers have turned up at different times.
You can see what I believe is the sampling period by looking at [[MediaWiki:monobook.js]]; it's currently 5000, and hasn't been lower than 3000.
However, I can't see why Irukandji jelly fish would be so high. It's not linked from anything, it's not in the news, and it's not even jellyfish season in Queensland. It could just be a statistical anomaly, or perhaps someone is deliberately messing with the stats?
I'm betting on anomaly.
That said... JonBenet Ramsay is interesting - it surged last night and with no link from the front page. Interesting to see how quickly our use (apparently) reacts to news stories. The sex-related ones - lists of porn stars etc - were consistently high and are now dropping; I wonder where they'll stabilise at.
Yes, that trend is interesting - why did they all show up so high then disappear. Was it a time of day thing? More stats required!! :)
I really hope we will end up with "last 24 hours" and "last 7 days" stats.
It looks like it's designed to work month-by-month - I'm not sure how reliable the figures would be at smaller "resolutions". Best ask Leon...
On 8/29/06, Andrew Gray shimgray@gmail.com wrote:
You can see what I believe is the sampling period by looking at [[MediaWiki:monobook.js]]; it's currently 5000, and hasn't been lower than 3000.
Heh, ok. So if we're seeing figures that end in 333, it presumably means the magic factor was like 4333 for a while or something. Yes, massive flaw in my logic, thanks for pointing it out ;)
I'm betting on anomaly.
Given the previous, then yes, that would be likely.
Steve
On 29/08/06, Steve Bennett stevagewp@gmail.com wrote:
On 8/29/06, Andrew Gray shimgray@gmail.com wrote:
You can see what I believe is the sampling period by looking at [[MediaWiki:monobook.js]]; it's currently 5000, and hasn't been lower than 3000.
Heh, ok. So if we're seeing figures that end in 333, it presumably means the magic factor was like 4333 for a while or something. Yes, massive flaw in my logic, thanks for pointing it out ;)
-333 has been there since the beginning; I'm guessing it's something to do with the algorithm used, but I dunno what. I'll ask Leon if I see him - I'm curious now!
On 8/30/06, Andrew Gray shimgray@gmail.com wrote:
-333 has been there since the beginning; I'm guessing it's something to do with the algorithm used, but I dunno what. I'll ask Leon if I see him - I'm curious now!
I'm guessing it's a division by three.
As in yesterday was the third day the script was running, and all the numbers ended in 0, 333 of 666 (or 667, I didn't look too closely). Today is the fourth day and all the numbers end in 0, 250, 500 or 750. Tomorrow they'll all end in 0, 200, 400, 600 or 800.
I would think that the script extrapolates the number of hits per page by multiplying the actual hits by the number used to set the sample frequency, 5000. It then divides by the number of days so far in the collection period (a month) to get hits per day.
Steve Bennett wrote:
On 8/29/06, Andrew Gray shimgray@gmail.com wrote:
You can see what I believe is the sampling period by looking at [[MediaWiki:monobook.js]]; it's currently 5000, and hasn't been lower than 3000.
Heh, ok. So if we're seeing figures that end in 333, it presumably means the magic factor was like 4333 for a while or something. Yes, massive flaw in my logic, thanks for pointing it out ;)
On day 3, they were all multiples of 1000/3
On day 4, they are all multiples of 1000/4...
-- Neil
On 8/30/06, Neil Harris neil@tonal.clara.co.uk wrote:
On day 3, they were all multiples of 1000/3
On day 4, they are all multiples of 1000/4...
I swear there's a pattern, but damned if I can figure it out...
Incidentally, Irukandji jellyfish is hanging in there, 6th position at +-19%. The top 10 is actually pretty stable, pluto down a bit and "list of big-bust models and performers" (what a mouthful) up.
The worst thing about stats like this is they really raise more questions than they give answers. Why Jeff Hardy? The next professional wrestler isn't until #36. But then again there are quite a few - it would seem that professional wrestling is actually more popular than pokémon. God help us.
More analysis needed.
Steve
On 8/30/06, Steve Bennett stevagewp@gmail.com wrote:
On 8/30/06, Neil Harris neil@tonal.clara.co.uk wrote:
On day 3, they were all multiples of 1000/3
On day 4, they are all multiples of 1000/4...
I swear there's a pattern, but damned if I can figure it out...
Incidentally, Irukandji jellyfish is hanging in there, 6th position at +-19%. The top 10 is actually pretty stable, pluto down a bit and "list of big-bust models and performers" (what a mouthful) up.
The worst thing about stats like this is they really raise more questions than they give answers. Why Jeff Hardy? The next professional wrestler isn't until #36.
we are top result on google for Jeff Hardy
we are not the top result for Kurt Angle
On 8/31/06, Steve Bennett stevagewp@gmail.com wrote:
On 8/30/06, Neil Harris neil@tonal.clara.co.uk wrote:
On day 3, they were all multiples of 1000/3
On day 4, they are all multiples of 1000/4...
I swear there's a pattern, but damned if I can figure it out...
See my post above.
What I said was that I would think that the script extrapolates the number of hits per page by multiplying the actual hits by the number used to set the sample frequency, 5000. It then divides by the number of days so far in the collection period (a month) to get hits per day.
That is, the script works out the number of hits per month, which is always going to be a multiple of 1000. Then it divides by the number of days in that month for which there is data.
Or popular with the readers, at least.
Interesting. I am surprised that "Irukandji jellyfish" came so high up on the list... is there some anomoly here?
...Interesting to see how quickly our use (apparently) reacts to news stories. The sex-related ones - lists of porn stars etc - were consistently high and are now dropping; I wonder where they'll stabilise at.
It really is a case of "be careful what you wish for", isn't it?
We were quite pleased over at en.wiktionary when Leon turned stats on there for us a couple of days previously, but the joy is now mixed with rue as we learn that among our most popular entries, so far, are consistently MILF, penis, fuck, masturbate, and breasts.
(It's true, though; clearly we'll all want to wait weeks or months before we can imagine that the stats in general aren't swamped by fixations du jour.)