Predicting the Major League Baseball Season

This paper attempts to predict the outcome of games from the 2012 Major League Baseball season. Sporting events are very important to many people, and professional leagues are worth billions of dollars. Baseball, in particular, is not only one of the most popular sports in the United States, but the vast amount of recorded data and statistics made publicly available also lends itself well to machine learning. Realizing that baseball games are quite noisy, in our prediction, we hope to predict games with sufficiently high accuracy and to unveil information about what makes a winning baseball team. A feature set was carefully chosen, and both classification and regression techniques were implemented. The performance of the algorithms were tested on a recent season and the results showed a small degree of success, but also confirmed the suspicion that baseball games are very hard to predict based off statistics alone.