Churn Prediction of Private Golf Club Membership Using Machine Learning

Private clubs such as golf clubs are looking for opportunities to prevent their members from canceling their services, subscription, or membership (Churning). An essential step in such strategies is proactively detecting Churning members. Utilizing technology for predictive analytics, precisely churn prediction, can be an advantage for golf clubs that aim to understand and maintain membership subscriptions. Hence, this study aims to create a model that predicts golf player membership churn by building and selecting the best model out of the five algorithms, which are Neural Network, Logistics Regression, Support Vector Machine, Boosted Decision Tree, and Decision Tree. The research methods are secondary data analysis and archival study from WHS.ph, which handles the database of all golfers in the Philippines. The results show that the best model that would identify the behavior patterns of a Filipino golfer that would churn out of their golf club membership is the Neural Networks model with the ANN nodes consisting of 19 Input nodes, 4 hidden nodes, and 1 output node. The ranking is based on the evaluation, which shows an accuracy of 88.4%%, precision is 89.1%, recall is 96.3%, F1 score is 92.6%, and AUC score is 90.4%.