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

[Solloby]

Thanks so much for the data Law, it is super appreciated.
I will run some Chi Tests on it later.
Now we have intralitter data, we should count some rereleases so we can compare pets across litters.

What we need to do is:
- Decide which years to look at for rereleases
- Identify the ID ranges for Dec 18
- Work out what pets to look at (not sure any Advent pet data will be useful in 2008 since they were separate adopts then)
- Gather data

Then we collate the data and release it all <3

[nickjr]

8D I'M EXCITED

I want to count the Advent 2009 rereleases for numbers because I feel that maybe Advent 2008 rereleases were sort of a test run for Tess (iirc I think she just wanted to do something nice for us on December 18, 2008?? will look for it later), so maybe she didn't plan the rerelease numbers suuuuper carefully. But to me, it would feel weird to just look at the Advent 2009 rereleases without looking at the Advent 2008 rereleases lol so... maybe count both??

Are we counting the Advent event pets as well or nah? We do have VR rarity change data on a bunch of them, I think (e.g. we know that the pre-July dogs went VR at roughly the same time, hence why suggestions to move the BA even further up on the current List make us cringe lol)

[Fizzlebee]

8D I'M EXCITED

I want to count the Advent 2009 rereleases for numbers because I feel that maybe Advent 2008 rereleases were sort of a test run for Tess (iirc I think she just wanted to do something nice for us on December 18, 2008?? will look for it later), so maybe she didn't plan the rerelease numbers suuuuper carefully. But to me, it would feel weird to just look at the Advent 2009 rereleases without looking at the Advent 2008 rereleases lol so... maybe count both??

Are we counting the Advent event pets as well or nah? We do have VR rarity change data on a bunch of them, I think (e.g. we know that the pre-July dogs went VR at roughly the same time, hence why suggestions to move the BA even further up on the current List make us cringe lol)

(Is also excited)

I think we should count Rerelease 08 because those pets will still count towards their overall numbers, even if only a small amount. Plus it would be interesting to see if there are any changes in rerelease ratios from year to year. As for Advent 08, we might still be able to get some useful data concerning rarity differences within litters, if there were any litters. (eg: tan vs grey mini husky) But I guess it would only be worth the time if we lack specific data.

[Law]

Hello again, it has been a while! I've written my tentative plans below.

I'll be sampling a random 10k from the following months/events:

• Advent 2008 (630,000-920,000)
• Rerelease 2008 (810,000-820,000) which won't be random I guess haha
• Rerelease 2009 (7,270,000-7,310,000)

I kind of also want to get 5k samples from each month from october 2008 and onwards. How does this all sound? :o

[nickjr]

Sounds good to me! Although maybe raise the sample size for October 2008 since... there was that 9-outcome litter on top of other regular litters lol or maybe the sample size is big enough anyway for that to not matter? idk

Would be interesting to see November 2008 because adoption limits were implemented after the litters were up for a few days AND I think Tess actually posted that there was a rare outcome for the card litter........?? I can't find a post from her from before the limits were added (but I feel like a lot of her early posts are missing) and I can't find any discussion about a rare card pup before the limits were added, but I did find a post AFTER the limits were added where she actually said the Joker had 1/132 chance of being adopted BEFORE the limits were added, and then there are other posts that tell me Tess gradually changed the Joker probabilities to something more reasonable AFTER the limits were added, and there's something about how someone managed to get three Jokers because they adopted 300 card dogs while others adopted 100 but got no Jokers lol so I really wonder what happened to all those "extra" card dogs.....

One more thing I think we should keep in mind is that the birthday pets were all available separately, so these are also pets for which I'm interested in seeing the rerelease data (along with the lime/ice rats, leopard rat, black rat, and tiger rat--were there others? can't remember).

lol yes may as well get a census of Rerelease 2008 lolol

[Law]

I think 5k is probably enough (october actually had less outcomes than in the previous months)? It's a huge drop from the 70k sample size we got for September '08, so it's hard to tell now how accurate the data will be xD

Perhaps I'll do 10k like the advent/rereleases

-----------------------------
Looking for help identifying pets!

I've been rewriting parts of the code, so now it takes all pets into account. So when the 2009 rereleases are counted, it can identify everything (even a random pet from June 2009, for example). The problem is, I don't have the names of all the pets. Currently, every pet has been assigned a numerical ID. It would be great if you guys could help me out by providing the pet names/litter names for these pets.

There are 875 pets in total from 2008-2009 (876 if counting the duplicate rat -- see below).I've split the pets into 9 sets of ~100, so it's easy to divide up the work. I can definitely use all the help I can get! [edit: removed] Actually, it would be easier to split it up by month/event, since the website (see below) is set up!

Please enter the pet/litter names into this spreadsheet: https://docs.google.com/spreadsheets/d/ ... sp=sharing

I've done part of July 2008, but everything else needs to be filled! The CS pet name guide is super helpful.

edit:
right now the stats page looks like this ;_;

edit edit:
I've gotten the data for christmas 2008 and the 2008 rerelease. Starting to work through 2009 rereleases, and then will do the monthly samples ^_^

-----------------------------
Random note:
This rat is in both July 2008 and Christmas 2008 in the archives


I put it under Christmas 2008 only :P

-----------------------------
CS statistics site WIP -- moved to separate topic: viewtopic.php?f=11&t=3510641&p=110696352#p110696352

[Law]

Bumping for visability~

I would like to release the advent/rerelease 2008 data asap, but getting the names comes first! Thank you to the anon user for helping out :D

[Solloby]

Sorry, been really busy lately.

I don't think the statistics can be released in that format.
Across-litter statistics are meaningless for months without adoption limits.
For same-litter statistics, we need to consider error margins, which is why we have been running Chi tests to verify if the numbers are different enough to be considered statistically significant or not.

For example, Pink Sorbet is 23% and Green Sorbet is 27% of the Sorbet litter based on our July count.
Looking at this data, a user might think Pink Sorbet is rarer than Green Sorbet, but that is not the case.
The Chi test results show that the difference is statistically insignificant; it was caused by randomness when adopting. A 2% error margin which is well within acceptable parameters. There is a 25% chance of any sorbet puppy adopted from a rerelease being either Sorbet, and that is what users need to know.

For the presentation of our results we probably need to run some statistics on them and present the results manually, and accompany them with explanations.

I am hoping we can calculate some approximate across-litter statistics for the months without adoption limits by using rerelease data, and I also have some calculations from the August data but it makes some big assumptions. We need the rerelease data to confirm whether or not those assumptions are correct before those calculations can be used.

Edit: Re pet names for the spreadsheet, can you use the August & September templates to get those names? Also are you planning to count all pets or just the ones without rarity change data?

[Law]

Even if pets didn't have adoption limits (or had extended releases, like the 2008 2nd gens and warrior cats), wouldn't random sampling still give us data about each outcome's probability of occurrence?

I guess I don't really understand why the Chi test is necessary. We know that there is going to be some margin of error, so obviously the %s aren't going to be exactly the same as the %s defined by the admins.
Pink sorbet: 23.26%
Blue sorbet: 24.90%
Yellow sorbet: 25%
Green sorbet: 26.83%
Even without the chi test, it is pretty clear that the sorbets have the same rarity. I think it's fine to publish the %s, if the users know that this is a result of a random sample.

Just curious, what assumptions are you using for the August data, and what are the assumptions for?

About the pet names: I could use them, but the numbers don't exactly line up to the #s I've assigned for each pet (which is the order from the archives). I could go through all the litters, but I'd prefer working on getting the site up and running quickly.

edit: I don't intend for the statistic site to be used as a major reference for trading, and certainly not for deciding rares list trading. The rarity list should use all the data we have, including results of the chi-test for insurance. For general purposes though, it would be cool to see things like the %s within litters for all months, or the chance of getting a sunpup from rerelease.

[nickjr]

Pink sorbet: 23.26%
Blue sorbet: 24.90%
Yellow sorbet: 25%
Green sorbet: 26.83%
Even without the chi test, it is pretty clear that the sorbets have the same rarity.

I don't think that's clear to people who have incorrect intuition about numbers and stats lol (aka probably a large number of people in the world...)

I'm the anon user who only filled in the Aug 08 balloon litter lol I plan on filling in everything soon, hopefully tonight or tomorrow night

also MAJOR SPARKLY EYES MY DREAMS ARE COMING TRUE @ your stats site
except I definitely agree with Sol that we can't present the data like that OTL at least not without an accompanying chi-square test column or something that shows "even though more green sorbets got adopted than pink sorbets in their original release, they're not significantly different in rarity"

numbers are powerful, but a lot of people think that they're more powerful/absolute than they actually are

edit: lol yikes, you can hear my distrust of humanity in this post @___@;; a better point is brought up by Solloby below -points down-