PoE 2 servers are restarting in: .
They should be back up in approximately .Python Scripts (40-quality recipes, party XP share)
|
merged the alternate version for now, thanks for your work.
" that sounds great actually, though i'll probably still add a (then quite higher) limit/warning/whatever to the exhaustive search in order to avoid ruining peoples session, just in case. |
|
|
I was wondering about the distribution of quality. Let's assume weapon/armor where the range seems to be 5-20 (except for stuff picked up from a strongbox or similar). Is the distribution even across the range, or is there some kind of skew? I'd like to know if anyone has reliable data or information about this?
|
|
|
interesting question, but i have no idea where we could get such data without collecting it first-hand. even if we would look through peoples stashes etc. it won't be representative e.g. some people might only hoard higher quality numbers.
that said we probably dont need that many datapoints to get a decent estimate of the distribution, writing them down while playing is no work either so i guess i'll do that. i just hope the ilvl is not relevant in this case... |
|
|
Very neat scripts, thanks for sharing! Looking at the github it doesn't seem to have a license stated, are you planning to add a license to the code to outline you're allowed usages? Interested in maybe using the 40% in some personal projects if your licensing allows.
|
|
" thanks for pointing that out, i added a simple MIT license (its basically open source now) and i hope & guess databeaver is okay with that. i consolidated the two approaches and added a proper argument parser, though some stuff is still missing. it would be nice if you could take a look at it sometime databeaver, i'd like to remove the redundant reach40_fast.py when you approve of that. no news on the quality distribution yet, i didn't get to play much recently. if someone reading this is up for posting his or her findings (in any format you like, just please make sure it is not distorted e.g. because you already sold all items with low-level quality values in the past), i'd gladly plot and analyze them. |
|
" MIT license is fine with me. " I'll check it out tomorrow (if I remember). " My quality gems in prophecy (haven't sold any yet): 3 5 6 6 6 7 7 9 11 11 13 14 17 18 The 3% gem must have dropped from a strongbox with "contained items have 3% quality", because qualities below 5% can't otherwise drop. The sample is fairly small, but if others have similar data they can be combined. |
|
" Awesome, already working on using it in some of my test scripts. "
Spoiler
iLVL: 27, Q: 13
iLVL: 11, Q: 9 iLVL: 40, Q: 8 iLVL: 28, Q: 13 iLVL: 40, Q: 5 iLVL: 5, Q: 17 iLVL: 34, Q: 7 iLVL: 34, Q: 17 iLVL: 34, Q: 16 iLVL: 33, Q: 6 iLVL: 16, Q: 10 iLVL: 36, Q: 13 iLVL: 35, Q: 18 iLVL: 29, Q: 13 iLVL: 34, Q: 8 iLVL: 40, Q: 10 iLVL: 46, Q: 6 iLVL: 33, Q: 11 iLVL: 19, Q: 5 iLVL: 58, Q: 13 iLVL: 47, Q: 13 iLVL: 55, Q: 8 iLVL: 58, Q: 5 iLVL: 35, Q: 6 iLVL: 41, Q: 5 iLVL: 67, Q: 10 iLVL: 45, Q: 9 iLVL: 63, Q: 13 iLVL: 35, Q: 6 iLVL: 65, Q: 15 iLVL: 61, Q: 6 iLVL: 34, Q: 8 iLVL: 35, Q: 10 iLVL: 56, Q: 9 iLVL: 66, Q: 11 iLVL: 19, Q: 9 iLVL: 55, Q: 9 iLVL: 33, Q: 15 iLVL: 50, Q: 8 iLVL: 61, Q: 9 iLVL: 49, Q: 10 iLVL: 48, Q: 7 iLVL: 56, Q: 11 iLVL: 60, Q: 6 iLVL: 36, Q: 20 iLVL: 55, Q: 13 iLVL: 66, Q: 15 iLVL: 33, Q: 9 iLVL: 64, Q: 2 iLVL: 61, Q: 16 iLVL: 15, Q: 8 iLVL: 46, Q: 5 iLVL: 42, Q: 5 iLVL: 30, Q: 5 iLVL: 36, Q: 6 iLVL: 38, Q: 13 iLVL: 46, Q: 16 iLVL: 31, Q: 16 iLVL: 64, Q: 7 iLVL: 34, Q: 7 iLVL: 55, Q: 11 iLVL: 44, Q: 16 iLVL: 46, Q: 6 iLVL: 63, Q: 5 iLVL: 55, Q: 11 iLVL: 67, Q: 5 iLVL: 67, Q: 11 iLVL: 53, Q: 13 Here is a Flask tab full (right before selling): Qualities: [13, 9, 8, 13, 5, 17, 7, 17, 16, 6, 10, 13, 18, 13, 8, 10, 6, 11, 5, 13, 13, 8, 5, 6, 5, 10, 9, 13, 6, 15, 6, 8, 10, 9, 11, 9, 9, 15, 8, 9, 10, 7, 11, 6, 20, 13, 15, 9, 2, 16, 8, 5, 5, 5, 6, 13, 16, 16, 7, 7, 11, 16, 6, 5, 11, 5, 11, 13] I've included a full list of the Quality and the item level inside the spoiler if you needed that data as well. Last edited by unv_annihilator#6579 on Jun 12, 2016, 2:39:03 PM
|
|
" nice, i'm curious to see what you're gonna do with it :) let us know if you need any further adjustments of the script. " wooo data! thanks a lot! got right to it. Detailed results:
Spoiler
i removed the single 2% quality occurance as an outlier and i suspect there might have been another strongbox with 13% quality of content in play, because there are unusually many of those. could just be a coincidence though. the correlation between quality value and item level is -0.105. correlation is covariance scaled to [-1, 1] and shows how dependend two variables are from each other. being close to 0 indicates that they are not dependend, at most with a slightly negative correlation: as one value goes up, the other goes down. if anything, one would expect the quality to go up with higher item level so it seems likely that there is simply no correlation between the two. the bars represent the number of occurences, the colored lines are polynomials that were fit onto the data.  using the maximum number of theoretically viable degrees for the polynomials (14) shows clear overfitting for higher values.  looking at polynomials of degree 1 to 5 shows a linear trend.  this linear polynomial is most likely the most accurate (considering they probably didn't implement the distribution as a very complex function). the function is p(x)=-0.4779x+10.1618. note that this could be distorted if the 13% quality stems mostly from a strongbox. Summary about quality distribution: note that the sample data is limited so theres obviously no guarantee that the following is true, but it still seems likely when looking at the data: 1. the item level has no influence on the quality values of dropped flasks. 2. high quality values are less frequent than lower quality values, this decrease is linear. i'll take a separate look at gems sometime, but i suspect its the same rule for every type of item. Last edited by Laoshra#3960 on Jun 14, 2016, 8:17:09 AM
|
|
" Nice finds. Something I will mention is that most of the 20% qualities I get I vendor without tossing in the stash. Only missed 1 in this set. So could probably add 4-5 20% but probably would still be in-line with what you found. If I get a good set of gem quality I'll send them over as well. |
|
