|
|
| Author |
Thread Statistics | Show CCP posts - 0 post(s) |
ShadowRat
   |
Posted - 2007.03.27 08:33:00 -
[1]
Edited by: ShadowRat on 27/03/2007 08:37:30
Lets start with multi formula:
succes rate= 1* (1+item in use) * (1+ enc.lev*0.02) * (1+ related science I.lev*0.02)* (1+ related science II.lev*0.02)* decryptor_mod
Multiply is resonable, but you can use 'add' too and results is not much diferent (form 67% to 85% at best config)
bpc:round_down(base_runs/30)+decryptor_mod+ little_lottery(after 1.4 patch)
me:-4+round_down(oryginal_me/30)+decryptor_mod
pe:-4+round_down(oryginal_pe/30)+decryptor_mod
If bpc runs after formula is 0 ... invention fail too, so this my be 'lost thing' that ppl dont consider in calculation.
Example:
Encryption lev = 4 -> 1.08 modifier
science I = 4 -> 1.08 modifier
science II = 4 -> 1.08 modifier
best descryptor =1.3
item in use = 0.07 (best!?!) -> 1.07 modifier
That give you around 75% of succes.
When you dont use items only skils .. succes rate is around 25%
Seams resonable for me... any commnts so far?
---------------------------------
Have a nice day
|
ShadowRat
|
Posted - 2007.03.28 07:45:00 -
[2]
Edited by: ShadowRat on 28/03/2007 08:25:03
Ok - lets see some points:
 Quote:
The formula for runs was already correctly* posted in this thread and looks nothing like the above. Try getting MaxRunsInputBP, MaxRunsOutputBP(no, the two are nearly never the same) and ActualRunsInputBPC in there.
If correctly ...
MaxRunsInputBP: that is nice, but what is MaxRUn for BPO? Yes bpc makes for max time in lab... but where? In station lab or pos? If you use 'standard station lab' BPC in your formula you got 100% but when you use from pos 133%. You must put another formula for pos bpc and some new modifiers :P, but why pos bpc must be trethed in onther way?This is little silly and too complicated in checking and implementing.
bpc_output: round_down(base_runs/30)+decryptor_mod + little_lottery(after 1.4 patch)
Is simply linear and continous function (without lottery). Easy to implement and there by only mistake in number '30' that is 'constans' number for all BPC (if?). Our job is found of what this number is. 30? 40? 80? But when you look at ppl's number they got more/less from 1run bpc 4 (1/30+4).... from 100run bpc 7 (100/30 +4).. from 300run 14 (300/30 +4)... You see? 30 is almost there...
Good side:
- no difference in ship/mod/rig/weapon.. etc BPC... Yes. Why there must be diference? When eve engine prevent you to make ship BPC more in run then modules... Simply less run input less you get.
-simple
Now 'little lotery' - that can be happen when you got next roll in luck and put you extra runs.
The me and pe. If you can make any number (in theory) in me and pe constans modifier here can be huge... 30 is only example.
Quote:
Furthermore, most important comment, the whole formula has a max(1;formula) around it (*that was missing from the formula post). There are no jobs failing cause of 0 runs.
When you got 0 run BPC ... yes you faill. No you win :) but you got nothing... so you fail anyway. And you count this as 'bad luck' in succes_rate... but this is a different story.
Quote:
About chance, i have a good guess about the formula and its wildly different from yours ;). I do need more statistics for any real hint of validity though.
At the end anyone use own formula. More or less acurate.
-----------------
Have nice day.
|
ShadowRat
|
Posted - 2007.03.29 12:38:00 -
[3]
Acording to inputs by ppl here succes rate formula can looks like this:
succes_rate= (1+encryption.lev/10)*(1+related_science1.lev/10)*(1+related_science2.lev/10)*(1+(meta_item.lev*5+5)/10)*modyfier_decryptor
There is low skill weight.. so you put 1.01,1.02, ...1.05 at max lev for 3 skils...
If item is put into invention you got 1.05,1.10,...1.25 modyfier at last from meta level of item.
If there is no item this part is eqal to 1.
Modyfier_decryptor is like we know, in example 1.3 for best descryptor.
So best chance is:
1.05*1.05*1.05*1.25*1.3=> 88%
if no item and decryptor
1.05*1.05*1.05=> 15%-16%
Some ppl imputs in this thread:
In order:
encryptionscience Iscience IImeta_levdecryptorSUCCESmetaleverror_rangePPL_SUCCES_REPORTED
1.041.041.041.251.382.790447.178.50%
1.041.041.041.151.368.16716827.7061%
1.041.041.041.21.375.478784316.650%
1.041.041.041.251.382.7904416.6100%
1.041.041.041.051.353.543936033.333%
1.041.041.041.21.148.482048311.155.55%
1.011.011.011103.0301noitem9.142%
1.031.031.031.21.370.4654123472%
1.041.041.041.251.382.7904411.177%
1.041.041.041.251.382.790441080%
As you see imputs are in range of this function +/- error range
I wonder why dev put here 0.4 decryptor... This sugest another few more paramets that help to get succes rate.
So.. use best items (if exist) and best decryptor.. :P
-------------
Have Nice Day
P.S.
No point in max function :P If you got 0 BPC you fail :P
|
ShadowRat
|
Posted - 2007.03.30 11:42:00 -
[4]
Another pices of puzzle:
What can be additiona modifier for succes rate?
-used interface ship/data/tuner and it influence on 'base' chance :)
(most resonable cus you use it)
-equpment level in station/pos
-standing to corp/fraction operating labortory
-race for items types
..
-weather in Iceland. :D
Any ideas?
|
ShadowRat
|
Posted - 2007.03.31 18:54:00 -
[5]
My next few coins:
ship_clas t1max_bpc t2mac_bpc
frigates 30 10
crusers 15 10
b.crusers 10 10
b.ships 10 10?
dreds 1 ?
carrier 1 (low class) ?
titan 1? (due a time) ?
Dominix BPC is nice example and give GOOD clue :)
You dont put it into invention... because:
This job cannot result in an output blueprint. The reason might be that the metalevel of your input blueprint is too high or that there are no items in the same metagroup with a higher metalevel.
So :) metalevel is included in formula... for shure:)
Have nice day :)
|
ShadowRat
|
Posted - 2007.04.18 12:01:00 -
[6]
Edited by: ShadowRat on 18/04/2007 11:59:04
Ello all again.
More my little coins ... or as some one sugest more clouds ;)
(all data is after 1.4.1 and to last database reset)
(I use ships data like base for my formulas)
-----------------------------------
Last time I start with simply multiplication. Now I know more about this formula from your posts infromation (thx for all :). Problem for develop this, is that ppl send diferent numbers... sometimes very diferent , and we dont know for shure if this is a problem of luck, ppl diversion or missunderstanding, ccp changes and more. SO?! We dont look exactly on NUMBERS but at correlation between NUMBERS.
Lets look at diferent side on this formula:
Succes ratio= (skils_modifier)*(interface_type)+(decryptor_mod-1)/10+(meta_item_mod-1)/10
ALL YOU PUT HERE AND GET BACK->IF YOU GET NUMBERS FROM 1.00 to 1.99 THIS SHOW YOUR SUCCESS RATE
1.00 is 0%
1.99 is 99%
LESS OR MORE NUMBERS ARE TRUNCATED TO 1 OR 1.99! (If it is too much for you get better computer)
ok.. now why ...and litle explain.
I.
skils_modifier=(1+encryption_skil*0.03)*(1+science_lev_1*0.01)*(1+science_lev_2*0.01)
I stil wonder if ther is * or +, but this is no mater because of specific build of this formula (read more)
Next: One tread sugest that only encryption is taken to it... hmm... maybe :) but this not mach all.
For this ppl:
skils_modifier=1+encryption_skil*0.04
But this part may looks like:
skils_modifier=1+MAX_level(encryption_skil,scienec_1,science2)*0.04
Whatever you put there for walidation in this formula you get numbers form 1.03 (first) to 1.32 or 1.04 to 1.2 (at last part)
You decide if ccp put here more variation or less.. And you can develop own part.
II.
Interface type: read more ;) we back here
III.
+(decryptor_mod-1)/10
+ here is esentional. I wonder why ccp put 0.4. This is HUGE in multiplication reduction of luck... now ppl start to send post with success rate with this decrypton.
SO? I look back at invention... and EUREKA :)
IN simply word you can get form this part:
0.4->-6%
0.8->-2%
1.0->+0%
1.1->+1%
1.3->+3%
This is simply true for ships, because of low variations.
In future this can be look like
+(decryptor_mod-1)/10*interfacetype_mod2... but we now stop at this....
IV.
+(meta_item_mod-1)/10
I use item_meta lev in similiar way like decryptor
Let put here someting like:
for no item ->0.95
0/or no metalev item->1
1->1.05
2->1.10
3->1.15
4->1.20
True for ships... and for modules it can be +(meta_item_mod-1)/10*interfacetype_mod3
but who knows :)
Now back to II.
Interface.... interface ... great mystery :) but we can outsmart this.
You must put here base_variation for interface type.... not only for ship, data or tuning... You must take diference from race/faction too!
After looking and this ther can be simply 12 base_modificatores... ;) Dont wory I show oyu how oyu can find it...
Example....
Lets use
(1+encryption_skil*0.04)*(interface_type)+(decryptor_mod-1)/10+(meta_item_mod-1)/10
Encsci1sci2metadecinterface deviation
1.041.031.030.951xsc11.18752.08%
1.041.041.0411.3xsc21.22502.50%
1.041.021.020.950.4xsc31.11112.78%
succesfrom+post
We dont know x.. but lets start witx x=1 for this kind of ship (yes covetors :P)
Encsci1sci2metadecinterface deviation
1.041.031.030.95111.15501.18752.08%
1.041.041.0411.311.19001.22502.50%
1.041.021.020.950.411.09501.11112.78%
succesfrom+post
close... put x=1.02
Encsci1sci2metadecinterface deviation
1.041.031.030.9511.021.17821.18752.08%
1.041.041.0411.31.021.21321.22502.50%
1.041.021.020.950.41.021.11821.11112.78%
succesfrom+post
Yes we are realy close :) And now we can use 1.02 IN THIS FORMULA FOR THIS SHIPS.. and preaty close you get good numbers...
put this modifiers into your ship interface...
|
ShadowRat
|
Posted - 2007.04.18 12:06:00 -
[7]
Now maybe my skils formula?
((1+encryption_skil*0.03)*(1+science_lev_1*0.01)*(1+science_lev_2*0.01))*(interface_type)+(decryptor_mod-1)/10+(meta_item_mod-1)/10
and x=1.08
Encsci1sci2metadecinterface deviation
1.041.031.030.9511.081.18661.18752.08%
1.041.041.0411.31.081.24491.22502.50%
1.041.021.020.950.41.081.10361.11112.78%
succesfrom+post
You got more variations... but as can you see formula match too.
SO?
How find yours numbers?
1. Choose you formula.
2. Put at last ONE invention succes rate of THE SAME INTERFACE!! More you put -> less mistakes you do.
3. Try get interface modyficator puting number that match posted succes rate (or your own) with deviation.
IF you get less deviation at one 'row' start think about this row like base...
After few trys you got good number.
4. Now you get YOUR VALID interface base multipler! USE IT FOR ALL SAME INTERFACE TYPE AND SAME FORMULA :)
What is all this for?
1. You can simply develop all kind of formula you can get into mind.
2. You never argue with other that eg. skils are more valid then meta_item or etc...
Simply rebuild formula and develop new interface base multipler.
3. YOU ARE NEVER BE NUMBERS SLAVE!
4. For more clauds into your mind :P
Thx again all for good thread and numbers.
P.S. Do at home match for modules :)
|
ShadowRat
|
Posted - 2007.05.22 07:30:00 -
[8]
Edited by: ShadowRat on 22/05/2007 07:29:01
Always yours 'point of view' depends where you sit...
If you saying that ppl can 'spare' milions in covetor bcause they dont put it to invention for hulk.. Let me show you some 'cloded' numbers.
If adding item with metalevel 1 can change your succes rate from (lets this be example only) by 1% to invention of modules like 80% of succes rate gives you 81%. This dont do much diference... but when you put this to ship invention -> from 10% you got 11% so... IT IS 10% MORE SUCCESS RATE then before... and this can be worth to do.
Have nice day.
|
ShadowRat
|
Posted - 2007.05.22 14:41:00 -
[9]
Edited by: ShadowRat on 22/05/2007 14:43:44
Bah... again:
(Maybe this time someone analyze it and share taughts)
(Diference from last my post with it is that I send interface
(modyficator and wirite is simply as I can :))
(BASIC VERSIOBN FORMULA!- USE ON YOUR OWN. FOR % USE SR%)
(WARNING: THIS IS ONLY MY CLOUDED MATH & HIPOTESE)
(I DO NOT PRETEND THAT IS THE ONLY ONE FORMULA)
(I JUST HAVE FUN WITH MATH - and... if this works :P)
(You can give me a beer)
------------------------------------------------------
SR= (1+(2*Enc + skill + skill2)/100)*I_TYPE + (DECR_MOD -1)/10 + (META_LEV_MOD -1)/10
SR%= (SR-1)*100%
DECR_MOD:
same as decryptor says .. from 0.4 to 1.3
I_TYPE: interface type
ship ->1.0
mod1 ->1.3
mod2 ->1.6
META_LEV_MOD:
no item ->1.00
1 meta ->1.05
2 meta ->1.10
3 meta ->1.15
4 meta ->1.20
If this is NOT easy .. I give up :P
For some reason and tuning purpose I use same formula but with this set of parameters:
I_TYPE:
ship ->1.03
mod1 ->1.3
mod2 ->1.58
META_LEV_MOD:
no item ->1
1 meta ->1.04
2 meta ->1.08
3 meta ->1.12
4 meta ->1.16
Have nice day...
|
|
|
| |
|