First of all, please excuse me if something similar was already started since ep4 update (any statistics made before are worthless, due to the massive changes the upgrade system went through then), a search didn't bring any interesting results.
As the title says, the goal here is to investigate the upgrade rates, and (I hope) decide some long-unanswered questions about them. I was doing this on my own before episode 4, but it needs a lot of trials to give accurate results, and it's rather impossible to do alone.
The protocol is very easy to follow : just download this attached file :
It contains an open office spreadsheet (it should work properly on excel too, only basic functions are involved, no macros).
And whenever you try an upgrade, write down the results in the grey cases. Don't change anything to the blue cases, it's where you'll see the results appear.
When you feel you have got accurate enough results or if you are simply fed up with upgrading, mail the sheet back to me so I can concentrate the results and post them back here.
If you wish to start again, then, fill the grey parts with 0es and go.
You can also just write the upgrades results like this 0 1 2 3 4 3 4 5 4 5 6 7 5 ... and send it here or in PM, I have a script to parse these.
Please if you do that, make sure to separate the numbers by spaces only, it's less hand correction for me before I feed my script.
Nobody forces you, it is obviously a lot of work if you only try an upgrade from time to time, but if you are a massive upgrades-addict, it might be worth it, for the good of all of us.
Thank you in advance!
>>>Results at the moment (click me)<<<
(last update : 26.08.12)
Also, if someone wants to suggest me better formulas (some are wrong as I see on the results, but I'm still waiting to find some time to fix them, so... ), feel free to help me.
I'd like some feedback about the relevance of the values. What would you want to see as calculated statistics? I especially wonder about Average penalty. For example, at 4>5 level, I have in record 1 +0 fail, 48 -1 fails and 1 -2 fail, which as a result give an average of -1. The possibility of failing +0 and -2 isn't obvious.
What should I do? Keep it as it is? Add the range of fails? Add the probability of each fail?
I'd like to thank all helpers, keep up the good work!
List of contributors
Edit - (01.06.11) Fixed the formulas for average penalty.
Edit - (03.06.11) Results updated.
Edit - (21.06.11) Results updated. More than 1000 cores counted!
Edit - (29.06.11) Results updated. More than 1500 cores now.
Edit - (11.07.11) Results updated. Feel free to submit bike upgrades and safeguarded upgrades. If you don't want to check each time how many safeguards disappeared, just add an "unknown fail penalty" column to the spreadsheet. Or simply give me the sequence and tell me when you were using SG. I'll upload an updated spreadsheet for that in a few days if people ask for it, but it seems noone uses this system anyway, so... ^^
Edit - (30.07.11) Results updated. Over 2700 cores! Some results for Average penalty are obviously wrong, I'll recheck the formulas soon.
Edit - (08.09.11) Results updated. Added some precision about how to provide results. We're over 3.7k cores now, 1.6k trials! Thank you all, keep up the good work!
How I wish that my data was here...
Edit - (17.09.11) Added google docs link.
Edit - (19.12.11) Finally updated the whole thing. More than 3k trials and 8k cores! O.O
Edit - (26.08.12) Big update today, 6.5k cores! That makes a total 6k upgrades and 14k cores! Also, added fields for superior cores, now waiting for you guys to fill these.