Basketball players performance measurement with algorithmic survival data analysis

Performance measurement is of paramount importance in the context of sports analytics. A great variety of data analysis methods has been exploited to this aim. All these proposals almost never include resorting to survival analysis techniques, although time-to-event data are suitable for addressing this issue. This work aims to identify the main achievements of a National Basketball Association player that affect the time it takes for him to exceed a given threshold of points. In order to identify nonlinear effects and possible interactions among the predictors, the analysis is carried out with machine learning methods, specifically survival trees and random survival forests.