Pokémon Stats ML Predictor
Can a neural network tell a Legendary from the stats alone?
A machine learning project using neural networks to analyze Pokémon base statistics and predict whether a Pokémon is Legendary, complete with confusion matrix visualization and interactive predictions.
Problem
Legendary Pokémon are rare and powerful — but can their status be predicted from raw stats alone, without any metadata? This project explored whether a neural network could learn that distinction.
Approach
I built and trained a neural network in Python using the full Pokémon stats dataset, experimenting with model architecture and hyperparameters to maximize classification accuracy.
- Preprocessed and normalized a dataset of 800+ Pokémon with 6 base stats each
- Designed and trained a multi-layer neural network classifier
- Evaluated model performance using a confusion matrix
- Built an interactive prediction interface for user-inputted stats
Outcome
The model achieved strong classification accuracy and could reliably identify Legendary Pokémon from stats alone. The confusion matrix revealed which types of Pokémon were most likely to be misclassified.