Fill up ESI data...

[jdegre]

Originally Posted by Mathayus

there's something confusing me...
if i got lets say 1k of a certain skill,and if u look to a chip that doubles your skill,i find chips worth around 10,11 ped tt...i got 1k,and with this chip i would get 2k..
but if i look a skill that i only have 5,6 or 10 for example and if i look for a chip with 1k skill its only worth around4,5 ped or not even that much....?| weird

No, it's not weird; what you have observed is exactly what those graphics try to show.

Basically, the "chip tt value" represents your skilling experience. It takes an equivalent skilling effort to fill an ESI from 0 to 10 ped, than from 10 ped to 20 ped.

However, the skill levels don't grow linearly. You go very fast from 0 to 2k, but it takes much much MUCH longer to go from 2k to 4k. This up-hill, is exactly what the graphics represent.

Therefore, from 0 to 1k you need tt=3.5ped (as you observed) but from 1k to 2k you need ~8.2 ped.

/jdegre.


PS: Kyl, I bet your last point is wrong:

618,53 6894 Kyl

Could you please double check it? I think it should be around 570 tt

[Kyl]

Originally Posted by jdegre

PS: Kyl, I bet your last point is wrong:

618,53 6894 Kyl

Could you please double check it? I think it should be around 570 tt

i guess you got me there

But honestly i am not sure about this one too but maybe you can help

Here is how i 'found' it:

I chipped out all my rifle in 4 steps with that results:

ESI TT used   	Level     	Skill lost	Skill remain   	ESI with Skill 	ESI back  	total back	Re Chip   	Lost TT   	Lost %    	ESI Needed
18,2      	7187      	52        	7135           	16,16          	0,24      	16,4      	7182      	1,8       	10,02%    	17,96     
337,38    	7135      	1068      	6067           	303,47         	0,18      	303,65    	7056      	33,73     	10,00%       	337,2     
314,27    	6067      	3615      	2452           	282,79         	0,04      	282,83    	5959      	31,44     	10,01%       	314,23    
71,42     	2452      	2451      	1              	16,11          	53,51     	69,62     	2351      	1,8       	10,05%       	17,91

all i could imagine was to calc the difference between start skill and re chip value, sum that up and subtract it from my starting skill value.

like that:

Level     	Re Chip   	Diff
7187      	7182      	5
7135      	7056      	79
6067      	5959      	108
2452      	2351      	101

7187 - 293 = 6894

Maybe this was way too nooob

[jdegre]

You cannot do that ("7187 - 293 = 6894"), since "skill levels" is a non-additive magnitude. However, "chip tt value" is additive (you can add, substract, take 10%, etc..)

From your data, we can get the following points:

2452 -> 17.91
6067 -> 332.14
7135 -> 669.34
7187 -> 687.30

Cheers,
/jdegre.

[Doer]

I busted out Matlab and played around with the data: i changed my mind about using log-log plots to fit it.

A 4th degree polynomial fit of a log-log plot is the lowest degree with relatively small residuals (slightly better than quadratic); it doesn't get any better with higher degrees. However, using such a fit equation to extrapolate shows a lot of divergence even by the last data points. On the other hand, trying to fit the entire data with up to ten degree polynomials without using logs is, as witte demonstrated, rather unsatisfactory.

It may be the breaking it up into segments and optimizing a polynomial fit for those segments would get better results.

I'm not sure that extrapolating above 8k skills is useful, in any case, because it's possible that the curve straightens out at some point. We simply don't know without some large data points (like a full tt ESI for example).

Edit: in short, nothing new to contribute -- just wanted to see for myself.

[Witte]

I got abit bored with my study (again) so thought lets solve this thing. In matlab, when you do:

plot (b, log(a./b)) (where a=ped and b= skill, so a./b = cost per skillpoint)

you get an almost linear relation for >1000. The coefficients are:

0.00054208(b)-6.2295. When you substract this from the plot above, you get all the leftover, which looks like:


Click to enlarge

These are pretty much the slowdowns and the speedups. I havnt realy analised these yet, but its seems the selected value is not correct, as it doesnt fit in the series (actam?).

Well back to work 4 me

[actam]

The data is got from chips, and is some case same tt did give me different skill-gain (starting always from 1):

#1 tt 1.00 skills from 1 to 1000 (just making the numbers for an example)
#2 tt 1.00 skills from 1 to 998

I did get the first one and max the similar TTs

but problem aroused:

1) to not know if the chips where max full TT filled
2) do not know the 3rd TT number... (tt 1.002 or tt 1.000), can't use sweat-method because I nataurally was in auction...
3) I do have data and double data of many but not on all TT values

I will insert other data when other chips will be disponible in auction hoping to get a better precision

[Witte]

Originally Posted by actam

The data is got from chips, and is some case same tt did give me different skill-gain (starting always from 1):

#1 tt 1.00 skills from 1 to 1000 (just making the numbers for an example)
#2 tt 1.00 skills from 1 to 998

I did get the first one and max the similar TTs

but problem aroused:

1) to not know if the chips where max full TT filled
2) do not know the 3rd TT number... (tt 1.002 or tt 1.000), can't use sweat-method because I nataurally was in auction...
3) I do have data and double data of many but not on all TT values

I will insert other data when other chips will be disponible in auction hoping to get a better precision

The small variations arnt realy a problem, but the point I am talking about, 36,74 3285 seems to be about 2 ped off. When I make it 38.74, it seems to fit the series allot better.

[actam]

Originally Posted by Witte

The small variations arnt realy a problem, but the point I am talking about, 36,74 3285 seems to be about 2 ped off. When I make it 38.74, it seems to fit the series allot better.

oh! might be a transcription error!

I did right down the TT and skill values as integers (3674/3285 for 36.74 -> 3285) and some times I do invert letters (and word)
So 3285 could have been 3825...
I did double check but some times sh*t happens

[Witte]

Originally Posted by actam

oh! might be a transcription error!

I did right down the TT and skill values as integers (3674/3285 for 36.74 -> 3285) and some times I do invert letters (and word)
So 3285 could have been 3825...
I did double check but some times sh*t happens

aah ok, so it could also be 37.64 ped. That also seem to fit better , but more data around that point is needed to know for sure.

[The187]

Good thread thx