Description

Built an MLB pitch predictor using Machine Learning to help coaches and players better strategize during games • Used a Kaggle data set consisting of over 2 million pitches from 2015 to 2018 • Utilized Scikit-Learn to build a random forest which classified 11 different MLB pitches with a 35% success rate • Feature engineered previous pitch and interaction variables

Demonstration of results

confusion matrix visualization

model performance

Future improvements

• test interaction terms
• implement a deep learning algorithm to test more potential models
• create a cross-validation script for assessing the accuracy of the model