PokeMeat

Problem Statement

What if Pokémon were edible? Our tool lets users input a Pokémon name and receive a predicted market price. Additionally, the user can also enter a market price and our tool will return a 5-member list of Pokémon priced around the input value. The market price is calculated using a machine learning model trained on attributes like HP, Attack, Defense, Special Stats, Speed, encounter rate, and more. These features loosely mirror how real-world meats are evaluated, such as beef being priced based on marbling, rarity, and consumer demand.

Basic Overview

To build our project, we used scikit-learn to create an MLP model to extrapolate prices from features of the first 151 Pokémon. We then created a full-stack project using FastAPI, React.ts, and Tailwind CSS to allow users to look up a Pokémon's predicted market price or reverse search the closest five Pokémon given a price per pound.

Challenges

We encountered challenges with the time constraint as we had a hard time translating Pokémon data to the meat market and debugging our MLP model. Translating the fictional logistics of the Pokémon universe to real-world meat market mechanics was difficult. We overcame this by simplifying our analysis to fewer variables. Additionally, many team members had to learn on the spot because we had low familiarity with API and back-end techniques. We collaborated on different laptops to see if the issue was in the code or environment.

Future Directions

This project can serve as a tool for imaginative worldbuilding, adding depth and dark humor to the fantastical universe of Pokémon by treating creatures as commodities. Beyond satire, it also opens the door to more complex modeling, where additional stats and their interdependencies could create a more detailed pricing algorithm. With further development, the framework could even be adapted to real-world scenarios, such as evaluating genetically modified meats, using biological parameters to predict value in the current market.