🧠 ML Pokémon Gen 1

Hellooo there, Pokémon Trainer! This repository contains my university project on Machine Learning, where I combined classic ML techniques with deep learning, using Pokémon as the core theme! 🐱‍🏍🎮

🎯 Project Objective

The aim of this project was to learn and apply basic Machine Learning concepts to a topic of our choice. Being a lifelong Pokémon fan, I decided to explore the world of Gen 1 Pokémon using ML techniques. Specifically, the classification was based on the Pokemon elemeental types (water,fire,grass,etc.)

📚 What I Did

🔹 Data Collection & Feature Extraction Collected images of Gen 1 Pokémon from five Kaggle datasets.

Applied Autoencoders and LDA to extract features.

🔹 Classical ML Models Used traditional ML algorithms:

• KNN

• SVM

• Linear Regression

• Naive Bayes

Evaluated performance using metrics:

• Accuracy

• Precision

• Recall

• F1 Score

🔹 Deep Learning Models Explored Neural Networks:

• MLP (Multi-Layer Perceptron)

• Basic CNN

• Deep CNN

Achieved 89% accuracy using Deep CNN.

📌 Notes

• The data cleaning notebook is not included in the repo.

• Each notebook is structured to be understandable and reproducible.

• The report includes detailed steps, challenges, and results.

🚀 Future Work

• Classify the images based on the evolution lines (Charmander, Charmeleon, Charizard)

• Classify the whole 151 Pokemon!

• Extend to other Pokémon generations!

📫 Contact

Feel free to reach out if you have any questions!!!