Onboarding Epoch V - Pokémon Type Classification

Team Epoch's solution to the Pokémon Type Classification Competition.

Getting started

Clone the repository and navigate to the project directory. Make sure Rye is installed on your machine and run:

rye sync

Alternatively, you can install the dependencies from requirements-dev.lock using the following command:

pip install -r requirements-dev.lock

pre-commit

This repository uses pre-commit for code quality checks and auto-formatting. To install the pre-commit hooks, run:

pre-commit install

To run the pre-commit checks on all files, run:

pre-commit run --all-files

Dataset

Dataset is available in the data/raw folder. It contains the following files:

Metadata

train_metadata.csv contains various details about each Pokémon, such as its name, type, species, height, weight, abilities, EV yield, catch rate, base friendship, base exp, growth rate.

CSV Descriptors:

• Pokemon: Name of the Pokémon.
• Type: One or dual type determining STAB (same-type attack bonus) and weaknesses or resistances to incoming attacks.
• Species: Identifies the Pokémon based on defining biological characteristics.
• Height: Height of each Pokémon.
• Weight: Weight of each Pokémon.
• Abilities: Special attributes aiding Pokémon in battle, introduced in Generation 3.
• EV Yield: Stats gained by defeating specific Pokémon.
• Catch Rate: Chances of catching a Pokémon with a Poké Ball.
• Base Friendship: Default friendship value when encountering a Pokémon.
• Base Exp: EXP yield when defeating a Pokémon at level 1.
• Growth Rate: Amount of EXP needed for leveling up.
• Egg Groups: Classification used in Pokémon breeding.
• Gender: Chance of Pokémon being male or female.
• Egg Cycles: Time unit for hatching Pokémon eggs.
• Base Stats (HP, Attack, Defense, Special Attack, Special Defense, Speed): Determine Pokémon strengths and weaknesses.

Images

Training images are located in the data/raw/train_images folder. Test images are located in the data/raw/test_images folder.

These images are in-game sprites of the Pokémon, sources from all mainline Pokémon games.

Submission

For making a submission you are expected to run inference on the data/raw/test_images and create a CSV file with the following columns:

id, Bug, Dark, Dragon, Electric, Fairy, Fighting, Fire, Flying, Ghost, Grass, Ground, Ice, Normal, Poison, Psychic, Rock, Steel, Water

An example submission file is provided in submission/example_submission.csv.