Update README of Ezreal Dataset (Patch 13.23)

[MiscellaneousStuff]

Update the section based on the following section and combine it with the existing Google Drive links.

Ezreal Dataset (Patch 13.23)
This dataset is comprised of 10,065 Ezreal early games from Patch 13.23 of League of Legends. The games were selected from a larger dataset of 57,667 replays across multiple regions (EUW1, EUN1, KR, LA1, NA1). The criteria used to filter the dataset were:
Only games where the Ezreal player was on the blue side
Games where the Ezreal player spent the least amount of time dead (empirically determined to be a good indicator of player skill and game quality)
Only the first 3 minutes of each game were included
The purpose of this dataset is to serve as a pilot study for training an AI agent to play Ezreal using supervised learning techniques. By focusing on a specific champion and the early game, the scope is narrowed to make the initial model development and training more manageable.
Dataset Details
Number of games: 10,065
Patch: 13.23
Champion: Ezreal
Game length: First 3 minutes of each game
Total frames: Approximately 5.4 million
The dataset includes detailed game state information extracted from the replays using the T_T-Pandoras-Box library. This information includes:
Champion data (position, health, mana, items, abilities, etc.)
Minion data
Monster data
Turret data
Missile data (i.e. projectiles like auto attacks and abilities)
The raw data is stored in SQLite3 database files, with one file per game. These raw files have undergone extensive data cleaning, normalization, and transformation to prepare them for use in machine learning models. The processed data is formatted into observations and actions suitable for supervised learning.