Pokémon TCG Rarity Predictor

Overview

A web app that predicts the rarity of a Pokémon card using deep learning. Input card features, and the model returns one of: Common, Uncommon, Rare, etc. This app is practical for Pokémon market design as it lets the model predict card rarity — a key factor in determining market value.

Software Architecture

Works Cited

1. Kaggle Dataset found at: Pokemon TCG All Cards 1999 - 2023
2. App Background image found at: Pinterest

Developer Configure

Install backend dependencies

pip install -r backend/requirements.txt

Train ML Model

cd backend
python3 train.py

After this, you should get two data files: model.pt and scalar.json

Run both servers in separate terminals

1. Open Frontend Server:

cd frontend
npm run dev

2. Open Backend Server:

cd backend
python3 api.py

3. Check your result by opening browser at localhost:5173 (default for vite)