Determine the Undervalued US Major League Baseball Players with Machine Learning

Abstract

Baseball is a sport of statistics. The industry has accumulated detailed offensive and defensive statistical data for over a century. Experience has shown that data analysis can give a competitive advantage compared to teams without using such analysis. In the last two decades, with the development of machine learning and artificial intelligence, we have had more advanced algorithms to analyze data in baseball. In the following research, we will run different ML models using sci-kit-learn and H2O on Colab, and the Caret package on RStudio to examine the datasets (hitting dataset and salary dataset) and determine the undervalued players by predicting the number of runs scored in the next year. We will compare machine learning regression algorithms and ensemble methods and give comprehensive explanations of the result. The suggestion of which model is superior in terms of prediction accuracy will be determined.