CS Rarity List [WIP] - status update on page 75

[Lya]

After having sorted my thoughts a bit, I know what I originally wanted to get at.

I won't explain much for now because I have no idea how to put it into words.

intended rarity = x
year of the rerelease = t
count of pet @ t-1 = y
count of pet @ t = z
number of pets released during the rerelease = a
correction ratio @ t = b

b = a/(z-y)

y ≙ x
z ≙ b

x = z*x/y

This gives a general idea of a pet's rarity, but it can in no way account for how many active users there are. We do however have data for over 5 years after 2012, so a general tendency should be visible.

[lil rascal]

Welcome back Lya

I agree with what you're saying, that's how I always thought re-releases work, tweaked each year, not a set variable. With the way the rarity system now takes into account only active users pets I think even Tess & Nick's calculations can be thrown, which I suspect is why we have the Mountain zebra back at VR for the first time in 7 years, more inactive users with the Mountain Zebra on their account came back online over the holidays, throwing the re-release calculations out. I'd say the same thing happens with the other pets but we only realise it when the re-release:active users pets calculations are actually thrown out enough to drop down in rarity

[Solloby]

So as far as I can tell, our next step is one of these options?
a) Someone gets Law's script working and we gather rerelease data
b) Someone writes a new script to gather rerelease data
c) We go back to manual counting to gather rerelease data

The rerelease data we need at minimum: 2008 - 2012
It would be nice to have 2013-2017, especially for UR pets.
Thoughts?

[Lya]

Personally, I would guess it might be faster to manually count.
Although, um.... *throws in mock-up spreadsheet that'd make counting faster* [old, unused mail without personal information, etc. etc.] (and google's default is to only see e-mails of people who were actually added to the sheet, so no worries, even when you're logged in - you appear as an anonymous animal)

PS: it wouldn't take more than 10 minutes to change the "relevant" column into a drop-down of pre-set names which get automatically counted

pps: it would take a bit more time to code it so you can choose main/sept/adv list (if wanted), relevant litters to the chosen list, and then relevant pets to the chosen litter (but it would serve to be able to count these things automatically and make error-spotting easier)

[DaDwarf]

Wow @ that spreadsheet! Cant there be made something that would automatically count those images? -doesnt know much about scripts-
Seeing em all underneath each other does look cool!


Doesnt each pet have their own image ID? Thats the same in every URL? Maybe that can be used in counting? Idk tho lol

[nickjr]

Ooh, I like that spreadsheet!

@DD So the thing Law made earlier automatically counted the images once &noitems=1 was added and the username tag thing was chopped off. Then all images of a pet would be identical. (If the username bit was left in, then the images of two identical pets would differ by the username bit lol) uuuggghh I understand the algorithm at a high level but don't know enough to implement it -crawls into hole-

And I was wondering the same as you about the pet IDs in images, but the thing is that there are only these versions of a pet image:

Code: Select all

[img]http://www.chickensmoothie.com/pet/4004656&trans=1&noitems=1.jpg[/img]

the version where you use the pet's unique ID (can change bg color/transparency, items)

Code: Select all

[img]http://static.chickensmoothie.com/pic.php?k=FC8308E4E0C1F519C78F2093F71DB7B6&trans=1[/img]

the version shown in pet groups and pet pages when the pet has no items on (can change bg color)

Code: Select all

[img]http://static.chickensmoothie.com/archive/image.php?k=FC8308E4E0C1F519C78F2093F71DB7B6&bg=cdcfd1[/img]

the archive image

It does look like there's some kind of ID for the pet itself (in this case, FC8308E4E0C1F519C78F2093F71DB7B6) but that ID and the unique pet ID never appear together...

[Solloby]

I'm not familiar with setting up Python scripts but if someone can get us started I can try to help with the coding.

[nickjr]

Spring break is coming up for me. I'll try to look into reproducing the script high-level overview/grabbing Law's script off of their pastebin link from somewhere in this thread and trying to figure out how to run it if Law can't get back to us soon, or if I can't figure that out then I'll mess with Lya's sheet. Have several appointments already scheduled over break but hopefully I can use my off days

The original gigantic discussion that started all of this was done December 2016 - January 2017 and it looks like more and more users over the past few months are voicing those same concerns but they just missed the boat and therefore thought there was no boat at all. And since this thread has been inactive, they don't see this thread, so I've been seeing a lot of borderline accusatory or outright accusatory questions about the status of the new List and I'm just feeling very, very tired.

Next steps:
> somehow get something up and running for pet counting
>> consideration: distinguish off-account from on-account
> get a census on all original releases and rereleases for the pets of interest (all 2008 non-extended release pets, just to start? other possible considerations)
>> ID ranges (~1.8 million pets, definitely want to spread that out) + notes
> once we have that data, then we can draw... whatever conclusions we need
>> can also hand data to Crestalia if her prof's still willing after all this time

The draft of the new List is currently with DD and Lya and is in dire need of more information because they only went off of the rarity changes and there's a whole slew of VR pets that have no rarity change data whatsoever.

We definitely want in-litter comparisons (e.g. one toxic is rarer than another).

Since we'll be getting census info, we should be able to directly compare pets from different litters and hope that we don't have to take inactivity into account. But we can only do this if we can exclude off-account pets, because there are a bunch of off-account rereleases adopted.

How does that look?

[Solloby]

Looks good Nickjr.

Some notes:
- Agree that off-account should be excluded
- We need to separate original release from rerelease numbers as original releases not having adoption limits drastically impacted their original numbers
- Nothing we can do about the Zebras being released throughout the year
- We may need to look at off-list VRs, and list VRs that are not 08s (e.g. Pink Shima BFW)

[Wookieinmashoo]

- We may need to look at off-list VRs, and list VRs that are not 08s (e.g. Pink Shima BFW)

I have been following this thread a bit and even though I don't think I've posted I can agree to this very strongly. I have been wondering when it would be brought up and looking at the rarity changes a while back, some of these pets do seem to be as rare if not more rare than some lower list pets. Though again, re-releases may skew that.