Interesting picture ;)

[DanC]

Dale,
Those are indeed some small koi.
I have received many requests for a trout pond site.
It seems that by popular demand, I must acquiesce.

It's ben a h*ll of a lot of work, but oh, so worthwhile. I'll announce it's location when I can get something put together.

dc

 

[Kris]

Hymy, thanks. When I saw that graph this morning, I thought wtf is up with that freehand red line? How come the red(edit:yellow) and blue are straight? I appreciate the 'ranting' explanation.

Russ. For some reason, your pic absolutely wiped me out. But I laugh hardest at the oddest things. Thanks.
Let's make it real, at least the blue line!

 

[Spoooon]

Maybe I should switch back to RC5...

 

[crisp]

OK, so here I am to give some more explanation about the graph wildhagen so kindly posted here
Offcourse I did give this explanation on the GoT forum, but I know that most of you guys will have some difficulty understanding Dutch

This is what I did:

For all 3 teams I took the total blocks submitted in the past three weeks. Let's call it week 1, 2 and 3.
I compared the difference between week 1 and 2, and between week 2 and 3. This I averaged to get an estimated rate of growth.
For DPC this was 3.2% per week, for AT -1.9% per week and for /. +0.3% per week.

Offcourse you can question wether a 3 weeks history is enough to predict a 3 months future, but problem was that if I was to go further into the past I had to go way back to the beginning of June to correct the DPC's rate for MF6 (we started saving for this one around May/June somewhere). This would just take too much time, so I decided to take only the past 3 weeks.

I used weekly totals to calculate the rates to correct for weekly flushes from some of our keyservers, and to correct for weekend influences.

Also for these 3 weeks I calculated the number of blocks retired to each team; this proved that DPC gets 7 times as much blocks retired to compared to AT and /. However in the total calculation this should not be a big factor.

Anyway using these factors and the average blocksubmission of the last 3 weeks I started calculating day by day the estimated score.

I also mentioned on GoT that when I initially calculated without 7 oktober stats the results were quite different from when I added the 7 oktober results (DPC had a rather big score for a Saturday). A longer term history is therefor necessary to get more accurate results.

Anyway, because the picture came out so nice I decided to post it anyway, just as an incentive for our team to keep up this rate! Not knowing it would pop up on your forum.

Hopefully it will also be an incentive to you to prevent us from obtaining first place before the end of this year

Happy cracking!

 

[Scraper]

Hymy,

Oke, you're right, but what i meant to say is that Wildhagen diddn't produce this pic. Sorry for that case of bad English

 

[Dale]

crisp thanks for the explanation

<<you can question wether a 3 weeks history is enough to predict a 3 months future>>

it is more than enough, but of course your rate of error will extend to your projections
IE
week 1 > week 2 + 10%
week 2 > week 3 - 10%
your method would show 0 growth, when in fact the growth would be closer to 4 1/5% than it would be 0
___________ above is an EXAMPLE only _______________

<<this would just take too much time>>

if it will help any , you havent had any real growth since Sept 1
including the MF
and most likely if you tried to project MF backwards your true growth is a negative
(although I didnt try that,yet)

hey, who is counting anyway ; )

..Dale

 

[hymy]

Damn Crisp,

I'm impressed. Thats how you document a graph. DAmn I'm just speechless. That was an incredible effort of prognostication.

I just have one last unanawered question.



<< Anyway using these factors and the average blocksubmission of the last 3 weeks I started calculating day by day the estimated score. >>




After you did this how did you plot those lines? I.e is it a xy scatter with the dots connected or just a simple line plot. Or did you regress the calculations.

 

[crisp]

hymy: thanks!
To answer your question; it's just an xy scatter with the dots connected, so I did not use regression.

By the way; I did some recalculation with 4 weeks history and came up with the date of December 22nd now since it seems we had a small dip after MF6. Also taking 4 weeks it seems that TA does have a positive growth of 0.7%; they just had a very bad week last week (an average of 1.35M blocks a day against 1.42 the week before that).

And Dale: you'd better do your math again because DPC definitely IS growing although it's an average of 1.6% per week when counted over 4 weeks history.

It seems 3 weeks history just isn't enough to do a good prediction; I think by mid December I will get it right