A Machine Learning Approach to Infield Shifts

For the past 10 weeks, with the assistance of the Wake Forest Baseball Analytics program and guidance from Computer Science professors at the University of California, Santa Cruz, I worked on a Machine Learning approach to optimize the defensive positioning of infields:

Abstract

The problem I want to solve is how to decrease the likelihood of an infield batted ball turning into a base hit, therefore simultaneously increasing its likelihood of turning into an out since those are the only 2 possible outcomes of an infield ball in play. I would like to solve this problem in order to minimize the number of runs, which are generated through base-hits, that the opposing team scores. This will help the Wake Forest Baseball team by decreasing their opponents’ runs scored, therefore increasing their own chances of winning baseball games and eventually the College World Series. To tackle this problem, I implemented Unsupervised Learning, specifically K-Means Clustering, and Supervised Learning, specifically Logistic Regression. Based on my results, I was able to decrease the likelihood of an infield batted ball turning into a base-hit by strategically shifting the position of defenders in the infield.

Full Report is attached via URL:

https://cutt.ly/Stgzna4

Full repository is attached via GitHub:

https://github.com/ayusharora99/WakeForest

Feel free to reach out to me via Twitter:

https://twitter.com/ayusharora1999