RevTpark / pokeDex

Hobby project created to learn more about ML and algorithms

PokeDex

Additional

• App
• Scrapper and ML

Tech Stack

• Jupyter Notebook
• Flask/Python
• HTML/CSS/JS

Functionalities

• Home Page allows to search pokemon based on Name, Type and Generation.
• Pokemon guesser using image works like a real pokedex. Upload an image to the guesser or take one. And it will guess the closest pokemon(Trained for the first 151 pokemon only).
• Pokemon guesser using text tries to guess a pokemon using a pokemon description provided to it. The description is matched to the closest pokedex entry from all generations(Trained for the first 151 pokemon only).
• Pokemon Team Strength Ranker is an algorithm that ranks a given team. How the algorithm works:
  • Overall the strength of the team depends on the type coverage as the stats can vary based on Level, IV's, Nature, etc. But type remains constant for all the teams.
  • Strength of the team is greater if the type coverage in terms of attack is wide.
  • Also higher the immunities to other types or resistant to other types better the strength.
  • If weakness can be covered by other team members, the strength increases.
  • If team is 4X weak to certain types, the strength decreases.