1 minute read

I’m excited to introduce my latest project: a website that predicts win rates in League of Legends based on champion compositions. Check out my project: https://gitformike.github.io/lol/

This project combines my interest in League of Legends with my passion for machine learning and web development. I wanted to create a tool that could provide insights into how different champion combinations might perform in a match.

Key Features

  • Win Rate Prediction: The core feature is the ability to predict the win rate of a given champion composition. By selecting champions for each team, the website uses a trained machine learning model to estimate the likelihood of each team winning.
  • Intuitive User Interface: The website provides a simple and easy-to-use interface for selecting champions and viewing predictions.

Technologies Used

This project utilizes the following technologies:

  • Next.js: A React framework for building server-side rendered and statically generated web applications, providing excellent performance and SEO.
  • TensorFlow.js: A JavaScript library for machine learning in the browser. This allowed me to train and deploy the prediction model directly on the client-side.
  • League of Legends Match Data API: I used a Riot Games API to gather historical match data, which was crucial for training the machine learning model.

How it Works

  1. Data Collection: I collected a large dataset of League of Legends match data from a public API. This data included information about the champions picked for each team and the outcome of the match.
  2. Model Training: I used TensorFlow.js to train a machine learning model(Auto ML) on this dataset. The model learned to identify patterns in champion compositions and correlate them with win rates.
  3. Prediction: When a user selects a champion composition on the website, the trained model makes a prediction about the win rate for each team.

Future Plans

I plan to continue improving this project by:

  • Improving the accuracy of the prediction model by using more data and exploring different model architectures.
  • Adding more features, such as champion statistics and team synergy analysis.
  • Improving the user interface and user experience.

I welcome any feedback or suggestions you may have! Please feel free to explore the website and let me know what you think.

Thank you for your interest!

Updated:

You may also enjoy