MLB next pitch prediction app with streamlit
Link to Web APP: https://thawing-citadel-06239.herokuapp.com/
The idea of this project was to predict the next pitch-type of an at-bat, given the in-game situation. For example, given that the count is 1-1, there are 2 outs, it is the bottom of the 4th inning, the run differential is 1 run, and Lucas Giolito is pitching....then the next pitch in an at-bat will be predicted as a 4 Seam Fastball.
You can learn more about the process here at my blog post (https://seanwsullivan1.wixsite.com/ssullivananalytics/post/pick-a-pitch-any-pitch) - note that for this model, the algorithm is Random Forest and not MLP Neural Network. This was my first time using Git as it is intended - not just as a host for my Jupyter Notebooks. It was a great learning experience.
The dataset is cleaned from Kaggle. I deployed the model using Streamlit and Heroku and recommend watching the following videos to learn how to do it step by step.
- https://www.youtube.com/watch?v=skpiLtEN3yk
- https://www.youtube.com/watch?v=mQ7rGcE766k (Especially helpful for using Git).