Nikolaj-K

"""
Code and prompt used in the video
https://youtu.be/xcE_0azawvM
"""

"""You'll get two tasks
First part:
Consider a joint distirbution of two random variables X_k with k in {a, b} (i.e. two random variables X_a and X_b),
over a finite outcome sets of size n_a and n_b.
Explain what the marginal distributions are.

Nikolaj-K

""" Script to the video at https://youtu.be/8w4jHN1LsnY """

===== Moving quantile series ===== === Preliminary - Exponential distribution example === Density: $p_\mu(x) := F'(x) = \frac{1}{\mu} {\mathrm e}^{-\frac{x}{\mu}}$

Nikolaj-K

"""
Proof prsented in the video

https://youtu.be/Ydyhe7KdRsk
"""

Reminder:

=== Theorem 1, the geometric sum and series formulae ===
$\sum_{k=0}^n a^k = \dfrac{a^{n+1}-1}{a-1}$

Nikolaj-K

"""
Code for the video at
https://youtu.be/CquQe7Y05VA

This script will be easy to generalize, also.
"""

from datetime import datetime
import imageio.v2 as imageio
from io import BytesIO

Nikolaj-K

# For an elaboration, see the description at the top of the follower-gain page at
# https://gist.github.com/Nikolaj-K/4aab20f7858267eded3f7e98dd1a980e


De.sol@desola__xn 3000.0 gained
3000.0 % SMART follower gain throughout 28 ranked days.
1 to 31 followers (dates 250326 and 250723)

Dmitry@dmitrysitskiy 2200.0 gained
2200.0 % SMART follower gain throughout 72 ranked days.

Nikolaj-K

The data below shows the follower gains of Somnia yappers, when looking at 
the first and last day they appeared on the 7-day Kaito leaderboard.

Example: @lovaniceth entered on March 23th and is there on July 23th also, in which time
he grew his following from 11991 to 13549, making for a gain of (13549/11991)-1 or 12.993%.

I post the dataset comprising 469 users twice, once sorted by longevety on the leaderboard, 
and once further below by gain.
I don't include users who were on the board for just one day or who lost subscribers (negative gain).

Nikolaj-K

#1 with 112 NFTs: 25 quills, 7 uprising, 25 grillz, 2 koda, 25 deed, 10 bambi, 10 pixcape, 8 demons
#2 with 140 NFTs: 1 quills, 2 uprising, 25 grillz, 50 bambi, 37 pixcape, 25 demons
#3 with 49 NFTs: 1 quills, 1 grillz, 2 koda, 39 deed, 1 bayc, 5 mayc
#4 with 27 NFTs: 25 quills, 2 pixcape
#5 with 51 NFTs: 1 quills, 1 uprising, 16 grillz, 1 koda, 12 deed, 9 bambi, 7 pixcape, 4 demons
#6 with 28 NFTs: 2 quills, 1 uprising, 2 grillz, 1 koda, 16 deed, 6 bambi
#7 with 24 NFTs: 1 quills, 1 uprising, 1 grillz, 1 koda, 14 deed, 2 bayc, 3 mayc, 1 bambi
#8 with 27 NFTs: 10 quills, 11 bambi, 6 pixcape
#9 with 19 NFTs: 1 koda, 11 deed, 2 bayc, 5 mayc
#10 with 12 NFTs: 12 quills

Nikolaj-K

DCT = {
    "aixbt@aixbt_agent": {
        "user_id": "1852674305517342720",
        "username": "aixbt_agent",
        "yaps_all": 29726.45,
        "yaps_l24h": 15.09,
        "yaps_l48h": 39.52,
        "yaps_l7d": 172.28,
        "yaps_l30d": 1933.83,
        "yaps_l3m": 4468.66,

Nikolaj-K

"""
Code described in the video

https://youtu.be/yyHd7BGGVp8

Note: A day after recording, I refactored the script to use a more continuous
random variable (no jump at full circle). I find this gives a nicer to read plot.
Secondly, I've also added a tiny `class State`, abstracting away the state indices
and conversions using string.ascii_uppercase.index.
"""

Nikolaj-K

import matplotlib.pyplot as plt

IMAGE_DIRPATH = r"your/path/to/lifespans.png"

PEOPLE = [
    ("Voltaire", 1694, 1778, "Philosopher"),
    ("Hume", 1711, 1776, "Philosopher"),
    ("Rousseau", 1712, 1778, "Philosopher"),
    ("Kant", 1724, 1804, "Philosopher"),
    ("Fichte", 1762, 1814, "Philosopher"),