Accurate Prediction of NBA Players’ Lifespan with Big Sports Data

Abstract

This study aims to figure out which factors can be a significant risk predictor for the lifespan of professional basketball players and to figure out which machine learning model can be used to predict NBA players’ lifespan. This paper explores the relationship between response variable American professional players’ lifespan and predictors including handness, total points earned during career (PTS), height, weight, position, and ethnicity using data from departed sportsmen who have worked in the National Basketball Association and the American Basketball Association. This study compares different machine learning and decides to use regression for the prediction. The cut-off date for death data collection is August 1, 2020. Overall, this analysis included and identified 920 deceased players. As a result, weight and PTS play the most crucial factor which affects NBA players’ lifespan. Besides, the range of weight is between 137.00 pounds to 284.00 pounds, with a standard deviation of 22.48, and has a mean weight of 199.03 pounds. Except birth periods before 1920 and from 1941 to 1950, the most massive players are more likely to die younger than the lightest players based on the descriptive finding. Similarly, Polynomial, Lasso, and Ridge regression analyses show a strong relationship between the predictor weight, PTS, and response lifespan. The mortality risk of more massive and high PTS players is significantly higher than lighter and lower PTS players. Understanding the potential risks of early mortality can help NBA players better plan their training sessions to improve their life quality.KeywordsLifespanRegressionMachine learningNBA playersBig sports data