This chapter examines the use of Random Forest, a nonparametric machine learning algorithm based on decision trees, for predicting rare events. A common challenge in binary classification is data imbalance, where one outcome occurs far less frequently than the other, such as loan defaults. Accurately predicting these rare events and evaluating model performance under such conditions poses special challenges. The chapter explores various resampling techniques that enhance prediction accuracy in imbalanced datasets. Additionally, it introduces the Precision-Recall curve, a key metric for assessing rare-event prediction. These methodologies are demonstrated using real-world survey data on child health in India, with all analyses conducted using the R programming language.