Inspiration
The inspiration for this project came from our deep passion for both machine learning and baseball. The Google Cloud MLB Hackathon provided an ideal platform for leveraging our skills in Machine learning and AI . We saw a chance to make impact in the world of sports analytics.
What it does
Our project uses machine leaning models to predict various aspects of MLB performance, while also integrating the Gemini API for enhanced prediction explanation. This tool helps us provide insights and explain why a model makes specific predictions , making our project interpretable.
How we built it
We began by collecting and analyzing the MLB data , then built and trained multiple models ( including Linear and Logistic regression). After that we integrated our model with Gemini API allowing us to explain predictions effectively .
Challenges we ran into
A major challenge we faced was deploying our model due to the billing account issue on Google Cloud . We also struggled with fine tuning and training models with MLB data which pushed us to the edge of giving up multiple times . But we persevered because this was our first significant project , and we were determined not to quit .
Accomplishments that we're proud of
Despite not being able to deploy on Vertex AI , we are proud of our work integrating Gemini API with our model . The project now works as intended and we've learned so much about machine learning deployment , API integration and project management .
What we learned
We learned how to handle real-world challenges like deploying models , managing APIs , and overcoming technical limitations . The project taught us the value of persistence and importance of adapting to the constraints we faced .
What's next for MLB Talentforge
We plan to revisit the deployment aspect when we overcome the billing account issue . Additionally we aim to refine our model further and explore new ways to expand our analysis and insights into MLB world.
Log in or sign up for Devpost to join the conversation.