On 8/29/06, Andrew Gray <shimgray(a)gmail.com> wrote: 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? 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, geni <geniice(a)gmail.com> wrote: I'm sure it's place is cemented now that everyone who reads the mailing list has clicked on it ;)

On 8/29/06, Andrew Gray <shimgray(a)gmail.com> wrote: 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 ;) Given the previous, then yes, that would be likely. Steve

On 8/30/06, Andrew Gray <shimgray(a)gmail.com> wrote: 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. -- Stephen Bain stephen.bain(a)gmail.com

On 8/30/06, Neil Harris <neil(a)tonal.clara.co.uk> wrote: 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 8/31/06, Steve Bennett <stevagewp(a)gmail.com> wrote: 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. -- Stephen Bain stephen.bain(a)gmail.com