Brawl Coaching

Inspiration Our inspiration for Brawl Coaching came from our passion for gaming and the desire to help players improve their skills. We noticed that many players struggle to understand the nuances of Brawl Stars and how different brawlers and strategies can affect the outcome of matches. We aimed to create a tool that provides actionable insights based on historical match data and player performance.

What it does Brawl Coaching is a web application that leverages machine learning to analyze past Brawl Stars matches and provide tailored coaching recommendations. It predicts the potential outcomes of matches based on the selected brawlers, maps, and player strategies, helping users make informed decisions. The application also offers insights on how to improve gameplay by analyzing performance metrics and suggesting optimal team compositions.

How we built it We built Brawl Coaching using a combination of technologies. The frontend is developed using HTML and CSS, allowing for a visually appealing and easy to use user interface where players can select brawlers and view insights. The backend is powered by Node.js and Express, which interacts with the Brawl Stars API to fetch data and use TensorFlow.js to make predictions based on our trained models.

Challenges we ran into During the development process, we faced several challenges: Model Training: Training the machine learning model required fine-tuning to ensure accurate predictions, which involved experimenting with different algorithms and parameters. Brawl Stars API: The API for Brawl Stars is incredibly vague and has many issues, including CORS handling and IP locking, all of which we struggled with immensely while coding our project. Language Barrier: We had a hard time integrating the backend and the frontend of the project as one of us only knew HTML and CSS, so we weren't able to use react which led to some difficulties with the backend, including having to do all of the backend in javascript/express rather than Python/Express

Accomplishments that we're proud of We are proud of the machine learning model we developed, which achieves a high accuracy rate in predicting match outcomes based on historical data. Additionally, the seamless integration of the frontend and backend allows users to interact with the application effortlessly. We are also proud that we made the project look polished and the back-end work, given both of us had little to no experience at the beginning and we learned a lot.

What we learned Throughout this project, we learned valuable lessons about data science, machine learning, and user experience design. We gained hands-on experience with TensorFlow.js and the nuances of model training, and we improved our skills in API integration and React development. Additionally, we learned the importance of iterative design based on user feedback to enhance usability. We also learnt about the process of webscraping and how to properly utilize API while integrating a front and back end network simultaneously.

What's next for Brawl Coaching In the future, we plan to expand Brawl Coaching by incorporating more advanced features, such as: Real-time Match Analysis: Implementing features that provide live feedback during matches. Enhanced Prediction Models: Continuously improving our machine learning models with more data and refining our algorithms. Community Features: Adding a social component where users can share strategies, tips, and insights with each other. Mobile Optimization: Developing a mobile-friendly version of the app to reach a wider audience of players.