Identifying key predictor variables of putting surface disruption using machine learning

Abstract The increase in golfing rounds since 2020 has led to more acute putting surface disruption (PSD) on golf course playing surfaces. Specifically, the concentrated PSD at the hole location is a concern for the game of golf and playing surface management. This study was designed to investigate the contribution of variables such as golf footwear tread type, walker, putting surface characteristics, and prevailing weather conditions on PSD, as measured by visual traffic damage and ball roll distance, near a simulated hole location. A random forest machine learning approach was applied to a dataset of 19 golf footwear trials conducted at the Cornell University Turfgrass and Landscape Research Center in Ithaca from 2019 to 2023. Results indicate that visual traffic damage could be only moderately predicted ( R 2 = 0.48), with the most important factors being the individual walker and tread type, whereas ball roll distance was unable to be reliably predicted ( R 2 = 0.16). These findings highlight the need for further investigation into variables that could better inform maintenance strategies that mitigate PSD and, thus, improve the golfer experience.