Clarify how per-slot and per-shop probability is used

[adamthedog]

Currently, the website is unclear in how the game uses the probabilities listed to fill slots with animals.
The per-slot probability implies that the game uses a discrete random number for every slot and uses the weights to pick an animal to put in that slot.
As an example to illustrate for others, here's what a per-slot random weighted choice might look like in Python:

import numpy
def number_of_slots(turn):
    if turn <= 4: return 3
    elif 5 <= turn <= 8: return 4
    elif 9 <= turn: return 5

animals = ("A", "B", "C", "D")
animal_weights = (
    # A,  B,   C,   D
    (.3, .3,  .4,  .0),  # turn 1
    (.2, .2, .35, .25),  # turn 2
    (.1, .1,  .4,  .4)   # turn 3
)

r = numpy.random.default_rng()
turn = X  # whatever turn it is currently
animals_in_shop = list(r.choice(a=animals, size=number_of_slots(turn), p=animal_weights[turn]))

where the var animals_in_shop is the list of final animals in the shop.
However, I'm not sure (and I'm sure others have encountered the same issue) what the per-shop weight means. Is that a cumulative probability of each animal appearing while factoring in all slots plus rolls? Is it a weight in some separate random choice that isn't discrete with respect to individual shop slots?

I'd really appreciate a clarification in the comments here, and maybe eventually the information can be added to the website.
As an additional aside/request, where is the prob. data from? Is it calculated from testing, or is it somehow mined from the game?
Thanks!

[bencoveney]

Hi @adamthedog

The probabilities at the moment are calculated as follows:

• For each turn, count the number of pets/food available.
• All pets/food are assumed to be equally likely, I do no have any weights (or the information available to calculate/look up weights, if they exist).
• The likelyhood that each pet/food will appear in a shop and slots is then calculated, given the number of slots available for that turn.

I understand that the sloth is a rare easter egg which would effectively have a very low weight if percentages were calculated that way, however at the time I was not able to find any information on the appearance probability so I just excluded it from the calculations.

The code that calculates those probabilities is here:
https://github.com/bencoveney/super-auto-pets-db/blob/main/src/db/populateProbabilities.ts

I exclude the sloth from calculations here:
https://github.com/bencoveney/super-auto-pets-db/blob/main/src/db/populateProbabilities.ts#L26

Either way, having a bit more background information easily available on how these values are populated sounds like a good idea to me!

[adamthedog]

Ah, so the per-animal probabilities in the game are actually just equal chances and the website's probabilities listed are a function of the equal chances and amount of slots available, I see!

I guess the sloth is likely calculated by a simple random chance that succeeds or fails to replace a slot with a sloth when the server generates the shop items.

I may make a PR that adds a clarification regarding this clarification of what the stats mean.
Thanks!