Predicting Rare Events Using Ordinal Regression - From Purchases to Diseases

Learn how ordinal regression can improve rare event probability estimation in a conference talk from Data Con LA 2023. PhD Candidate Greg Faletto from USC's Data Sciences and Operations department demonstrates practical applications for predicting uncommon occurrences like purchase conversions from click ads or rare disease diagnoses. Explore step-by-step methods, examine real-world data examples, and gain access to implementation code while covering key concepts including decision boundaries, logistic regression, intermediate outcomes, proportional odds models, and ridge penalties. Master techniques for handling class imbalance challenges through detailed explanations and hands-on examples in this 41-minute technical presentation.