Tool: Determine the teams vaults based on decks played the round (written in R)

[kaempfer13]

Heya,

I've been wasting a lot of time writing a tool that will allow us to determine the other teams vault based on the decks they played. I used the language R (https://www.r-project.org/), which is commonly used by us statistic students, but is probably mostly unknown otherwise. Its free, so you might try it if you want, but I can just do all the stuff with it myself and post the results.
Imo it allows you to do a lot of stuff with just a few lines of code, but its probably quite confusing to get into.

Links to vault tools:
aether
air
darkness
death
earth
entropy
fire
life
light
time
water


Anyways, here's the code (comments start with #):

#list of pillars to be excluded from vault
pillars<-unlist(strsplit("4sa 4vc 50u 52g 542 55k 576 58o 5aa 5bs 5de 5f0 5gi 5i4 5jm 5l8 5mq 5oc 5pu 5rg 5t2 5uk 606 61o 63a 6qq 6ts 6ve 710 72i 744 75m 778 78q 7ac 7bu 7dg 7f2 7gk 7i6 7jo 7la 7ms 7oe 7q0 7ri 7t4 7um 808 81q 8pj 8pk 8pl 8pm 8pn 8po 8pp 8pq 8pr 8ps 8pt 8pu"," "))
#insert the deck codes of the decks played the round between the ""
a<-""
b<-""
c<-""
d<-""
e<-""
# f merges all the decks cards
f<-paste(a,b,c,d,e)
# pillarvault contains all cards of the decks used in the round, with each card stored as an element of the charactervector pillarvault
pillarvault<-unlist(strsplit(f," "))
#vault only stores the nonpillar cards (mark cards are not excluded)
vault<-pillarvault[!pillarvault %in% pillars]
# the function unupper (defined below) turns all cards into their unupped version like they are in vault
vault<-unupper(vault)
#sort the cards for better overview
vault<-sort(vault)
#pointcounter (currently undefined) works similiar to unupper, assigning each card in vault their point cost and adding it for all known cards
points<-pointcounter(vault)
# change the code back to the format used for deck codes
vault<-paste(vault, collapse = ' ')
# print vault to console
vault

#define function unupper
unupper <- function(vault){
#list of upped card codes
upped<-unlist(strsplit("6rj 6rk 6rl 6rr 6rs 6tt 6tu 6tv 6u0 6u1 6u2 6u3 6u4 6u5 6u6 6u7 6u8 6u9 6ua 6ug 6uq 711 712 713 714 715 716 717 718 719 71a 71b 71c 71d 71e 71f 71k 71u 745 746 747 748 749 74a 74b 74c 74d 74e 74f 74g 74h 74i 74j 74o 752 779 77a 77b 77c 77d 77e 77f 77g 77h 77i 77j 77k 77l 77m 77s 786 7ad 7ae 7af 7ag 7ah 7ai 7aj 7ak 7al 7am 7an 7ao 7ap 7b0 7ba 7dh 7di 7dj 7dk 7dl 7dm 7dn 7do 7dp 7dq 7dr 7ds 7du 7e4 7ee 7gl 7gm 7gn 7go 7gp 7gq 7gr 7gs 7gt 7gu 7gv 7h0 7h1 7h2 7h3 7h8 7hi 7jp 7jq 7jr 7js 7jt 7ju 7jv 7k0 7k1 7k2 7k3 7k4 7k5 7k6 7kc 7km 7mt 7mu 7mv 7n0 7n1 7n2 7n3 7n4 7n5 7n6 7n7 7n8 7n9 7ng 7nq 7q1 7q2 7q3 7q4 7q5 7q6 7q7 7q8 7q9 7qa 7qb 7qc 7qd 7qe 7qk 7qu 7t5 7t6 7t7 7t8 7t9 7ta 7tb 7tc 7td 7te 7tf 7tg 7th 7ti 7to 7u2 809 80a 80b 80c 80d 80e 80f 80g 80h 80i 80j 80k 80l 80m 80s 816"," "))
#list of unupped card codes
unupped<-unlist(strsplit("4t3 4t4 4t5 4tb 4tc 4vd 4ve 4vf 4vg 4vh 4vi 4vj 4vk 4vl 4vm 4vn 4vo 4vp 4vq 500 50a 52h 52i 52j 52k 52l 52m 52n 52o 52p 52q 52r 52s 52t 52u 52v 534 53e 55l 55m 55n 55o 55p 55q 55r 55s 55t 55u 55v 560 561 562 563 568 56i 58p 58q 58r 58s 58t 58u 58v 590 591 592 593 594 595 596 59c 59m 5bt 5bu 5bv 5c0 5c1 5c2 5c3 5c4 5c5 5c6 5c7 5c8 5c9 5cg 5cq 5f1 5f2 5f3 5f4 5f5 5f6 5f7 5f8 5f9 5fa 5fb 5fc 5fe 5fk 5fu 5i5 5i6 5i7 5i8 5i9 5ia 5ib 5ic 5id 5ie 5if 5ig 5ih 5ii 5ij 5io 5j2 5l9 5la 5lb 5lc 5ld 5le 5lf 5lg 5lh 5li 5lj 5lk 5ll 5lm 5ls 5m6 5od 5oe 5of 5og 5oh 5oi 5oj 5ok 5ol 5om 5on 5oo 5op 5p0 5pa 5rh 5ri 5rj 5rk 5rl 5rm 5rn 5ro 5rp 5rq 5rr 5rs 5rt 5ru 5s4 5se 5ul 5um 5un 5uo 5up 5uq 5ur 5us 5ut 5uu 5uv 5v0 5v1 5v2 5v8 5vi 61p 61q 61r 61s 61t 61u 61v 620 621 622 623 624 625 626 62c 62m"," "))
# recognize each upped card
for (i in 1:length(vault)){
if (vault[i] %in% upped){
# change the upped card to the unupped version by finding the card code in the same position in the unuuped vector like it used to be in the upped vector
vault[i]<-unupped[upped==vault[i]]}}
# output
return(vault)
}

#define function pointcounter
pointcounter <- function(vault){
#list number of points for the card codes
pointlist<-c(30,40,40,35,30,50,60,65,70,70,50,150,30,120,30,90,30,80,30,80,45,55,30,70,30,50,30,30,90,50,90,130,30,60,30,30,80,120,80,40,70,40,60,40,75,90,70,30,100,40,70,60,30,40,110,30,60,40,60,30,40,90,120,45,50,100,60,80,70,75,120,60,60,30,40,70,30,40,50,60,30,60,30,40,60,100,30,55,30,110,50,130,100,60,90,60,80,100,60,70,60,50,60,40,60,70,60,80,75,60,60,30,80,30,45,50,80,160,40,60,50,30,30,55,50,60,50,130,30,60,40,110,75,80,45,70,65,120,60,60,40,30,85,60,80,110,70,110,120,30,30,50,90,70,55,30,80,130,30,30,50,40,70,85,60,65,90,40,90,140,50,120,90,50,30,60,30,110,30,30,60,30,120,80,60,150,70,50,65,40,150,30,60,90,30,30,100)
#list of unupped card codes
unupped<-unlist(strsplit("4t3 4t4 4t5 4tb 4tc 4vd 4ve 4vf 4vg 4vh 4vi 4vj 4vk 4vl 4vm 4vn 4vo 4vp 4vq 500 50a 52h 52i 52j 52k 52l 52m 52n 52o 52p 52q 52r 52s 52t 52u 52v 534 53e 55l 55m 55n 55o 55p 55q 55r 55s 55t 55u 55v 560 561 562 563 568 56i 58p 58q 58r 58s 58t 58u 58v 590 591 592 593 594 595 596 59c 59m 5bt 5bu 5bv 5c0 5c1 5c2 5c3 5c4 5c5 5c6 5c7 5c8 5c9 5cg 5cq 5f1 5f2 5f3 5f4 5f5 5f6 5f7 5f8 5f9 5fa 5fb 5fc 5fe 5fk 5fu 5i5 5i6 5i7 5i8 5i9 5ia 5ib 5ic 5id 5ie 5if 5ig 5ih 5ii 5ij 5io 5j2 5l9 5la 5lb 5lc 5ld 5le 5lf 5lg 5lh 5li 5lj 5lk 5ll 5lm 5ls 5m6 5od 5oe 5of 5og 5oh 5oi 5oj 5ok 5ol 5om 5on 5oo 5op 5p0 5pa 5rh 5ri 5rj 5rk 5rl 5rm 5rn 5ro 5rp 5rq 5rr 5rs 5rt 5ru 5s4 5se 5ul 5um 5un 5uo 5up 5uq 5ur 5us 5ut 5uu 5uv 5v0 5v1 5v2 5v8 5vi 61p 61q 61r 61s 61t 61u 61v 620 621 622 623 624 625 626 62c 62m"," "))
# no cards calculated yet mean that 0 points are used yet
points<-0
# find the point value of each card and add it
for (i in 1:length(vault)){
# add value of new card
points<-points+pointlist[unupped==vault[i]]}
# output
return(points)
}

I noticed that this might have been more trouble than its worth, gotta do all the theorycrafting for this round tomorrow, but at least I finished it. Applying it to each element should be quick, so I think I'll finish that today or at least early tomorrow (Im all outta time right now though).

Edit: Added function pointcounter, fixed some typos

[kaempfer13]

New function compare: It figures out what new cards were used this round and adds them to the vault. it achieves that by comparing our previous knowledge of their vault with the vault that makes up the current round. Keep in mind that some of these cards might be discarded already!

I couldnt be bothered to comment the function, unless you are very familiar with the programing language it will look like gibberish. If you request so I can explain it in detail, but I wont do it if noone actually cares so long as it works.

compare<- function(vault1,vault2){
cards1<-rle(vault1)$values
cards2<-rle(vault2)$values
count1<-rle(vault1)$lengths
count2<-rle(vault2)$lengths
mergecards<-c(cards1,cards2[!cards2 %in% cards1])
totalcounts<-numeric()
reccurcards<-cards2[cards2 %in% cards1]
for (j in 1:length(cards1)){if (cards1[j] %in% reccurcards){
if (count1[j]>count2[cards2==cards1[j]]) totalcounts[j]<-count1[j] else totalcounts[j]<-count2[cards2==cards1[j]]
} else totalcounts[j]<-count1[j]
}
totalcounts<-c(totalcounts,count2[!cards2 %in% cards1])
mergedvault<-""
for (y in 1:length(totalcounts)){ mergedvault<-c(mergedvault,rep(mergecards[y],totalcounts[y]))
}
mergedvault<-sort(mergedvault)
mergedvault<-paste(mergedvault, collapse = ' ')
# output
return(mergedvault)
}

usage in context (c and d contain the deckcodes of the previous mergedvault):
#example for the round 1 aether vault
c<-"4su 4su 4su 4su 4su 4su 52i 52i 52i 52i 52i 52j 52j 52r 52r 52r 52r 58q 58q 592 592 592 592 592 592 5f4 5f4 5f4 5f4 5f4 5f4 5f6 5f6 5f6 5f6 5fc 5fc 5fc 5li 5li 5lm 5lm 5lm 5lm 5lm 5lm 5rp 5rp 5rp 5rp 5rp 5rp"
d<-"61q 61q 61q 61q 61q 61q 61q 61q 61q 61q 61q 61q 61q 61q 61q 61q 61q 61t 61t 61t 61t 61t 61t 61t 61t 61t 61t 61t 61t 61u 61u 620 620 622 622 622 622 622 622 622 622 622 622 622 622 622 622 622 625 625 625 625"
e<-paste(c,d)
#format the vault so R can use it
vault2<-unlist(strsplit(e," "))
#the vault of this round
vault1<-unlist(strsplit(vault," "))
#use the function
mergevault<-compare(vault1,vault2)
#calculate the total points of cards revealed to have been in the vault
pointcounter(unlist(strsplit(mergevault," ")))

no i used as variable this time as I dont want to use [nobbq][/nobbq] all the time

[majofa]

You should do a gdoc like we have for our regular vaults.

And so much more can be ascertained by just assuming certain elements have between 18-24 or their key cards in their starting vault.

[kaempfer13]

I probably wont even manage to finish my work with my current method
I'd start making it around Tuesday next week; any ideas how to easily make it so that a list of cardcodes will translate into a number in the appropriate cells?
Maybe slam the whole deckcode of the known vault in a cell and add a function into the appropriate cells next to the card name that finds out how many times the corresponding cardcode is in that cell?

P.S.: Funny how we fight the first 5 elements in the alphabet this round. Removes some of the pressure to finish this without messing up the order.

[kaempfer13]

Vault tool gdocs for all other elements are out!

Now what's left to do is obtaining majofas E-mailadress (www3 can give me that too) and creating a tool that obtains the deckcodes from the rounds forumsection (based on the title you can determine teams, winner, roles) directly (does anyone know how Fippe made that tool or at least have the code for something similiar?).

Links to vault tools:
aether
air
darkness
death
earth
entropy
fire
life
light
time
water

How to use it:
the first sheet is called "vault that has been shown" it shows all cards that we can directly conclude from the decks played each round were definitely in the vault at some point. It does not factor in discards. Pls dont edit it for your assumption as to how the vault really looks like, I am not going to write those cell formulas again, so whenever you ruin one I'll revert the gdoc until before you edited it. I'll edit it once each round, by editing cell Q63 to match the mergevault of the team.
For speculations as to how the vault really looks/looked like, please make copies of it in which you are free to replace the cell formulas with any number of cards that you think make sense. I made one of those in the sheet called "presumed vault"; likely we're better of replacing it with a copy of sheet1 for it calculates for the current element correctly, while presumed vault assumes aether as element for now (also some of the numbers will be close to the truth).
I also provided sheets for assumptions about discards/salvage, but I will not edit it until I have the tool, that directly copies deck codes from each round for the given element (or its opponents for salvage).

Also for the sake of better visibility what all should I put in spoilers in the elements threads? each "deckcode" individually, sum up the rounds result in one spoiler and put the rounds standing of mergevault in another single big spoiler or other ideas.
Also i kinda want to have the individual decks as well, including opponents element and w/l for a lot of that information is lost in the current state of my tool.

[kaempfer13]

The tool on my PC is now almost completely automated, I only need to find an elegant way to not swap the decks if the loser (as ruled by the WMs (mostly because of illegal decks) posted his deck first (if someone is posted as the winner/loser in the results and countdown thread, it will look for the deck in the 1st/2nd post in the corresponding thread for the match  regardless of whether they really posted in that order or not).
For now I'll just post the results of the tool with that error still in there (at least it will be easier to find those decks, because of the 50% in-element rule).
Once I fixed that bit, I will also have all the discardable/salvageable cards for all elements (No point posting them inaccurately now).
Will be extra-tough to fix majofas wrong deck, as its posted nowhere close to the first 2 posts, I suppose I'll have to fix it manually (no 1button click to get all data for  /  )

[worldwideweb3]

majo can edit his post, dw.

[kaempfer13]

Most vaulttool issues are fixed and both the sheets and the new Forumthreads I made are fully updated. Will add Discards+salvage to the sheets sometime way later today.

I think I leave the old threads as they are, as they dont contain any information that you couldnt deduce from the new ones and it's much easier to get an overview of the new ones.

When I add discards/salvage to the sheets they should become far more relevant to our strategy, so it would be good if majofa could give me his Email, so he can actually see them (It was his idea after all). Just say that you are mafa (if you're it) and I'll share the sheets with your E-mailaccount.

[kaempfer13]

Vault tools are functional, you can watch the impact of adding discards/salvage on the discardsvault.

How to use: In the worksheet Presumed_discards all the decks to be discarded from are shown, but fully unupped and without marks/pillars.
In the 3rd column you can fill in the card code of the cards you think the team discarded after that match. Make sure that the code matches the decks code, otherwise we others might be confused (also the Code has to be unupped, else it wont change the sheet discardsvault). You are encouraged to do that, I dont want to do everything alone in those sheets!
The sheet Presumed_salvage works the same way and influences both salvagevault and discardsvault.

In each workbook there's a worksheet called Presumed vault or Presumed_vault. You can edit those freely, as they are not part of the calculations.

Current limitations: - For salvage it only adds those cards to salvagevault (and thus discardsvault), if fewer copies of that card have been played in any round (as it has no way of telling whether they have been played after they were salvaged or before then atm)
- It only calculates salvage for entries in the 4th column of Presumed_salvage and the 3rd column of Presumed_discards (If there's no entry for a deck, it wont impact calculations; you can also post completely unrelated deckcodes, which only makes sense for vaultdiscards, so I'd advise against it, I have no failsafe that makes sure that the original deckcode and the one entered by you match)
- It only takes the maximum amount of a card in any round as the basis for the copycount before calculating salvage/discards.
This has the following effects:
- It wont notice if in a previous round the copycount thats currently in Vault_that_has_been_shown has already been discarded and the Team still played a deck that contains those cards
- discardsvault can contain negative numbers

Errors that I'll probably fix after round 6:
For earth and entropy round 5 a deck has been swapped (as I entered one number incorrectly on my PC); this is less work to fix if I wait for Round 6 results.
salvagevault and discardsvault calculate in-element % and max copycount allowed incorrectly for non-
(gotta fix entries in 3 cells, I am sure one of you could do that, too)


Do we want to add the links to the sheets in our Titanpad?

Edit:
Also majofa, I granted you access to the sheets after seeing you online on the same time that the E-Mailadress I suspected was yours active in our official vault sheet; I already asked you twice for it in this thread, it would be nice if you communicated a little more with us.

[kaempfer13]

Do we care about the discard/salvage simulations at all? Given that we barely scrap by with our time and we would have to catchup on so many rounds, the answer is probably no.

[worldwideweb3]

Do we care about the discard/salvage simulations at all? Given that we barely scrap by with our time and we would have to catchup on so many rounds, the answer is probably no.

yh, dont bother about that. Just getting what decks teams use is enough