Predict Rising Stars in Sports: An Example of NBA Players

Abstract

Predicting rising stars is of public interest. Professional personnel search for rising sports stars is based on experience, which is usually highly subjective. Typically, players’ statistical data are used to make predictions, but rookie players lack abundant statistics, and some personal characteristics may not be reflected in the statistics. Alternatively, online discussions like tweets contain information about players, so the Text Sentiment Deep Prediction for Athletes Career (TSDPAC) algorithm was proposed to use tweets related to NBA players and sentiment analysis to predict whether a player is a rising star. The TSDPAC adopted deep learning techniques to construct the prediction model, which effectively predicted rising stars in the experiments and outperformed the compared algorithms.