Abstract by Joseph Leung
Incorporating Social Connections in Predicting Sport Teams
In the standard link-prediction problem one tries to determine the probability that a link will form in the future between two people or groups. Determining what is the liklihood is that an individual in one group may transition to another is an extension of this problem and has several applications. The one which we focus on considers how individuals transition between baseball teams. Assuming that an individual baseball player leaves a team, how might we predict which team the player will join next? We focus on different factors such as the success of a team and the evident performance of the player to see how they may affect a potential transition. In this setting we utilize machine learning algorithms such as Random Forrest to determine how accurate our model is, based on certain factors.