Pokémon Team Optimization

4 months ago

The author reminisces about their childhood love for Pokémon, including playing the games, watching the anime, and collecting merchandise.
As an adult, the author rediscovered Pokémon games but found themselves min-maxing their team, unlike their childhood approach of using a single strong Pokémon.
The author formulated the problem of optimizing a Pokémon team as a Mixed-Integer Problem (MIP), focusing on maximizing base stats and ensuring type resistances.
Key constraints include selecting 1-6 Pokémon, ensuring resistance to each type, and avoiding duplicate Pokémon.
The article explains linear optimization (LP) and the simplex algorithm, then discusses handling integer constraints with branch and bound.
Non-linear constraints for type resistances are addressed using auxiliary variables to enforce logical conditions.
The author implemented the solution using Python's PuLP library, leveraging a Kaggle dataset for Pokémon stats.
Optimal teams included legendary and pseudo-legendary Pokémon, with high base stats and strong type resistances.
Removing legendaries and pseudo-legendaries yielded teams with strong ordinary Pokémon, highlighting trade-offs between base stats and resistances.
The project demonstrates the application of operations research methods to game strategy, with potential for further customization.

Hasty Briefsbeta